I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
,
p
p
.
5
0
1
7
~
5
0
2
6
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
5
0
1
7
-
5
0
2
6
5017
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Cla
ss
ificatio
n of r
eg
io
na
l lang
ua
g
e
dia
lects using
con
v
o
lutiona
l
neura
l net
wo
rk a
nd multila
y
er
per
ceptron
F
a
hm
i B.
M
a
ra
s
a
bes
s
y
1
,
Dw
iza
Ria
na
2
,
M
uji
E
rna
wa
t
i
3
1
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
S
t
u
d
y
P
r
o
g
r
a
m
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
ma
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
U
n
i
v
e
r
si
t
a
s N
u
sa
M
a
n
d
i
r
i
,
J
a
k
a
r
t
a
,
I
n
d
o
n
e
si
a
2
D
o
c
t
o
r
a
l
P
r
o
g
r
a
m i
n
I
n
f
o
r
m
a
t
i
c
s
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
m
a
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
U
n
i
v
e
r
si
t
a
s N
u
sa
M
a
n
d
i
r
i
,
J
a
k
a
r
t
a
,
I
n
d
o
n
e
s
i
a
3
I
n
f
o
r
mat
i
c
s
S
t
u
d
y
P
r
o
g
r
a
m
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
ma
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
U
n
i
v
e
r
si
t
a
s Nu
s
a
M
a
n
d
i
r
i
,
Ja
k
a
r
t
a
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
4
,
2
0
2
4
R
ev
is
ed
Sep
11
,
2
0
2
5
Acc
ep
ted
Oct
16
,
2
0
2
5
Re
g
io
n
a
l
la
n
g
u
a
g
e
s
a
re
v
it
a
l
fo
r
c
o
m
m
u
n
ica
ti
o
n
a
n
d
p
re
se
rv
i
n
g
c
u
lt
u
ra
l
id
e
n
ti
t
y
,
sa
fe
g
u
a
rd
in
g
lo
c
a
l
h
e
rit
a
g
e
.
Ho
we
v
e
r,
g
l
o
b
a
li
z
a
t
io
n
a
n
d
m
o
d
e
rn
iza
ti
o
n
e
n
d
a
n
g
e
r
th
e
ir
e
x
i
ste
n
c
e
a
s
th
e
y
a
re
i
n
c
re
a
sin
g
ly
re
p
lac
e
d
b
y
n
a
ti
o
n
a
l
o
r
g
l
o
b
a
l
lan
g
u
a
g
e
s.
De
sp
it
e
p
ro
g
r
e
ss
in
d
iale
c
t
re
c
o
g
n
it
i
o
n
re
se
a
rc
h
,
p
a
rti
c
u
larly
fo
r
c
e
rtain
lan
g
u
a
g
e
s,
fu
rth
e
r
st
u
d
ies
a
re
n
e
e
d
e
d
to
im
p
r
o
v
e
m
o
d
e
l
p
e
rfo
rm
a
n
c
e
a
n
d
a
d
d
re
ss
les
s
-
re
p
re
se
n
ted
d
iale
c
ts,
i
n
c
lu
d
in
g
th
o
se
i
n
In
d
o
n
e
sia
.
T
h
is
st
u
d
y
e
n
h
a
n
c
e
s
a
c
u
sto
m
-
b
u
il
t
d
a
tas
e
t
fo
r
d
iale
c
t
re
c
o
g
n
i
ti
o
n
th
ro
u
g
h
t
h
e
a
p
p
li
c
a
ti
o
n
o
f
d
a
ta
a
u
g
m
e
n
tati
o
n
tec
h
n
i
q
u
e
s,
s
p
e
c
ifi
c
a
ll
y
a
d
d
in
g
n
o
ise
,
ti
m
e
stre
tch
i
n
g
,
a
n
d
p
it
c
h
sh
ift
i
n
g
.
Us
in
g
M
el
-
fre
q
u
e
n
c
y
c
e
p
stra
l
c
o
e
fficie
n
ts
(
M
F
CC
)
fo
r
fe
a
tu
re
e
x
trac
ti
o
n
,
it
e
v
a
lu
a
tes
th
e
p
e
rfo
r
m
a
n
c
e
o
f
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
(C
NN
)
a
n
d
m
u
lt
i
lay
e
r
p
e
rc
e
p
tr
o
n
(M
LP
)
i
n
c
las
sify
in
g
six
In
d
o
n
e
sia
n
d
iale
c
t
s.
Re
su
lt
s
in
d
ica
te
t
h
a
t
CNN
o
u
tp
e
rfo
rm
e
d
,
a
c
h
iev
in
g
9
7
.
9
2
%
a
c
c
u
ra
c
y
,
9
7
.
9
0
%
re
c
a
ll
,
9
7
.
9
7
%
p
re
c
isio
n
,
9
7
.
9
2
%
F1
-
sc
o
re
,
a
n
d
a
k
a
p
p
a
sc
o
re
o
f
9
7
.
4
9
%
wit
h
c
o
m
b
in
e
d
a
u
g
m
e
n
tati
o
n
tec
h
n
iq
u
e
s,
se
tt
in
g
a
fo
u
n
d
a
ti
o
n
f
o
r
fu
rt
h
e
r
re
se
a
rc
h
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dialec
t r
ec
o
g
n
itio
n
Mel
-
f
r
eq
u
e
n
cy
ce
p
s
tr
al
co
ef
f
icien
ts
Mu
ltil
ay
er
p
er
ce
p
tr
o
n
R
eg
io
n
al
lan
g
u
ag
e
d
ialec
ts
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Dwiza
R
ian
a
Do
cto
r
al
Pro
g
r
a
m
in
I
n
f
o
r
m
atics
,
Facu
lty
o
f
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
,
Un
iv
e
r
s
itas
Nu
s
a
Ma
n
d
ir
i
J
atiwar
in
g
in
R
o
ad
No
.
2
,
C
ip
in
an
g
Me
lay
u
,
Ma
k
asar
,
E
ast J
ak
ar
ta,
DKI
J
ak
ar
ta,
I
n
d
o
n
esia
E
m
ail: d
wiza
@
n
u
s
am
an
d
ir
i.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
Glo
b
a
liz
ati
o
n
h
as
h
a
d
a
s
i
g
n
if
i
ca
n
t i
m
p
a
ct
o
n
t
h
e
p
r
ese
r
v
at
io
n
o
f
r
e
g
i
o
n
al
la
n
g
u
a
g
es
a
n
d
l
o
ca
l
c
u
lt
u
r
al
id
e
n
t
iti
es,
p
a
r
ti
cu
la
r
l
y
i
n
I
n
d
o
n
esia
,
a
c
o
u
n
t
r
y
k
n
o
w
n
f
o
r
i
ts
r
i
ch
li
n
g
u
is
ti
c
a
n
d
cu
lt
u
r
al
d
iv
er
s
ity
.
Ac
co
r
d
i
n
g
t
o
a
r
e
p
o
r
t
b
y
UNE
SC
O
(
2
0
2
1
)
,
a
r
o
u
n
d
4
0
p
e
r
c
en
t o
f
t
h
e
w
o
r
l
d
'
s
la
n
g
u
ag
es
a
r
e
e
n
d
a
n
g
e
r
e
d
,
m
a
in
l
y
d
u
e
to
t
h
e
lac
k
o
f
in
te
r
g
e
n
e
r
at
io
n
al
t
r
a
n
s
m
is
s
i
o
n
.
A
s
im
i
la
r
t
r
e
n
d
is
e
v
i
d
e
n
t
in
I
n
d
o
n
esi
a
,
wh
er
e
t
h
e
u
s
e
o
f
r
e
g
i
o
n
al
la
n
g
u
ag
es
am
o
n
g
y
o
u
n
g
er
g
e
n
e
r
a
ti
o
n
s
i
s
in
cr
ea
s
in
g
l
y
b
ei
n
g
r
e
p
la
ce
d
b
y
I
n
d
o
n
es
ia
n
o
r
f
o
r
e
i
g
n
la
n
g
u
ag
es,
es
p
e
ci
all
y
E
n
g
l
is
h
,
i
n
d
ai
ly
ac
ti
v
it
ies
s
u
c
h
as
i
n
e
d
u
ca
ti
o
n
,
t
h
e
w
o
r
k
p
l
ac
e
,
an
d
s
o
c
ial
m
e
d
i
a.
T
h
is
s
h
i
f
t
r
ai
s
es
c
o
n
ce
r
n
s
a
b
o
u
t
th
e
p
o
te
n
ti
al
e
x
ti
n
cti
o
n
o
f
r
e
g
i
o
n
al
la
n
g
u
a
g
es
a
n
d
th
e
c
u
lt
u
r
a
l
v
al
u
es
e
m
b
e
d
d
e
d
wit
h
i
n
t
h
e
m
[
1
]
.
I
n
d
o
n
esia
r
an
k
s
as
th
e
f
o
u
r
th
m
o
s
t
p
o
p
u
lo
u
s
co
u
n
tr
y
in
t
h
e
wo
r
ld
,
with
ap
p
r
o
x
im
ately
2
7
3
m
illi
o
n
p
eo
p
le
s
p
r
ea
d
ac
r
o
s
s
1
7
,
5
0
8
is
lan
d
s
.
Mo
r
e
th
an
7
0
0
lan
g
u
ag
es
ar
e
s
p
o
k
en
th
r
o
u
g
h
o
u
t
th
e
ar
ch
ip
elag
o
,
h
ig
h
lig
h
tin
g
its
v
ast
lin
g
u
is
tic
d
iv
er
s
ity
.
Un
f
o
r
tu
n
ately
,
m
an
y
o
f
t
h
ese
lan
g
u
ag
es
ar
e
n
o
w
a
t
r
is
k
.
Acc
o
r
d
in
g
to
d
ata
f
r
o
m
E
th
n
o
lo
g
u
e,
4
4
0
lo
ca
l
lan
g
u
ag
es
in
I
n
d
o
n
esia
ar
e
class
if
ied
as
en
d
an
g
er
ed
,
an
d
1
2
h
a
v
e
b
ec
o
m
e
ex
tin
ct.
A
s
tu
d
y
o
f
9
8
lo
ca
l
lan
g
u
ag
es
f
o
u
n
d
t
h
at
n
ea
r
ly
h
al
f
ar
e
co
n
s
id
er
ed
e
n
d
an
g
er
ed
,
wh
ile
an
o
th
er
s
tu
d
y
r
ep
o
r
ted
th
at
7
1
o
u
t
o
f
1
5
1
lo
ca
l
lan
g
u
ag
es
h
av
e
f
ewe
r
th
an
1
0
0
,
0
0
0
s
p
ea
k
er
s
[
2
]
.
L
an
g
u
ag
e
is
clo
s
ely
ass
o
ciate
d
with
d
ialec
ts
an
d
a
cc
en
ts
,
all
o
f
wh
ich
ar
e
i
n
f
lu
e
n
ce
d
b
y
f
ac
to
r
s
s
u
ch
as
th
e
en
v
ir
o
n
m
e
n
t,
lan
g
u
ag
e
p
r
o
f
icien
c
y
,
an
d
s
o
cial
in
ter
a
ctio
n
[
3
]
.
A
d
ialec
t
is
a
v
ar
iat
io
n
o
f
a
lan
g
u
ag
e
s
p
o
k
e
n
b
y
a
s
p
ec
if
ic
g
r
o
u
p
o
f
s
p
ea
k
er
s
,
ch
ar
ac
ter
ized
b
y
u
n
iq
u
e
tr
aits
th
at
d
if
f
er
f
r
o
m
o
n
e
r
eg
io
n
to
an
o
th
er
[
4
]
.
Acc
o
r
d
in
g
to
th
e
s
tu
d
y
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
1
7
-
5
0
2
6
5018
th
e
d
y
n
am
ics
o
f
la
n
g
u
a
g
e
in
t
er
ac
tio
n
in
m
u
lticu
ltu
r
al
u
r
b
a
n
co
m
m
u
n
ities
,
ea
ch
la
n
g
u
a
g
e
p
o
s
s
ess
es
in
h
er
en
t
u
n
iq
u
e
n
ess
th
at
r
ef
lects
th
e
cu
ltu
r
al
id
en
tity
o
f
its
s
p
ea
k
er
s
.
Ad
d
itio
n
ally
,
ev
er
y
r
eg
io
n
ex
h
ib
its
d
is
tin
ctiv
e
lin
g
u
is
tic
ch
ar
ac
ter
is
tics
an
d
tr
ad
iti
o
n
s
,
wh
ich
ev
o
lv
e
th
r
o
u
g
h
co
n
tin
u
o
u
s
s
o
cial
in
ter
ac
tio
n
an
d
cu
ltu
r
al
b
len
d
in
g
am
o
n
g
d
iv
er
s
e
g
r
o
u
p
s
[
5
]
.
Fro
m
a
lin
g
u
is
tic
p
er
s
p
ec
tiv
e,
ap
p
r
o
ac
h
es
to
th
e
co
n
ce
p
t
o
f
d
ialec
t
ca
n
v
ar
y
am
o
n
g
lin
g
u
is
ts
.
Dialec
ts
en
co
m
p
ass
d
if
f
er
e
n
ce
s
n
o
t
o
n
ly
in
p
h
o
n
o
lo
g
y
,
lex
ico
n
,
o
r
g
r
am
m
ar
b
u
t
also
in
p
r
o
n
u
n
ciatio
n
an
d
ev
er
y
d
ay
lan
g
u
ag
e
u
s
e
[
6
]
.
T
h
e
r
ef
o
r
e
,
p
r
eser
v
in
g
r
e
g
io
n
al
lan
g
u
ag
es
is
e
s
s
en
tial
to
m
ain
tain
in
g
cu
ltu
r
al
id
en
tity
a
n
d
s
af
eg
u
a
r
d
in
g
lo
ca
l h
er
itag
e
.
Fo
r
f
o
r
ei
g
n
s
p
ea
k
er
s
lear
n
in
g
I
n
d
o
n
esia
n
,
co
m
m
u
n
icatio
n
s
u
cc
ess
is
o
f
ten
m
ea
s
u
r
ed
b
y
t
h
eir
ab
ilit
y
to
co
n
v
er
s
e
with
n
ativ
e
s
p
ea
k
er
s
.
Ho
wev
er
,
ch
allen
g
es
ar
is
e
wh
en
th
ey
in
ter
ac
t
d
ir
ec
tly
with
n
ativ
e
s
p
ea
k
er
s
d
u
e
to
v
ar
iatio
n
s
in
l
o
ca
l
d
iale
cts.
Acc
o
r
d
in
g
to
W
an
g
[
7
]
,
d
i
alec
ts
ar
e
also
d
if
f
icu
lt
to
b
e
ac
cu
r
ately
u
n
d
er
s
to
o
d
b
y
s
p
ee
ch
r
ec
o
g
n
itio
n
s
y
s
tem
s
b
ec
au
s
e
o
f
th
eir
u
n
iq
u
e
p
r
o
n
u
n
ciatio
n
,
v
o
ca
b
u
la
r
y
,
a
n
d
g
r
a
m
m
atica
l
s
tr
u
ctu
r
e.
T
o
ad
d
r
ess
th
ese
is
s
u
es,
d
ialec
t
r
ec
o
g
n
itio
n
h
as
in
cr
ea
s
in
g
ly
b
ee
n
in
te
g
r
ated
as
an
ess
en
t
ial
co
m
p
o
n
en
t
with
in
v
o
ice
r
ec
o
g
n
itio
n
tech
n
o
lo
g
y
,
en
ab
lin
g
s
y
s
tem
s
to
p
r
o
ce
s
s
r
eg
io
n
al
lin
g
u
is
tic
d
iv
er
s
ity
m
o
r
e
ef
f
ec
tiv
ely
[
8
]
.
Vo
ice
r
ec
o
g
n
itio
n
tech
n
o
lo
g
y
o
f
f
er
s
a
p
o
ten
tial
to
o
l
f
o
r
s
u
p
p
o
r
tin
g
th
e
p
r
eser
v
atio
n
an
d
u
n
d
er
s
tan
d
in
g
o
f
r
eg
io
n
al
lan
g
u
ag
es,
p
a
r
ticu
lar
ly
th
r
o
u
g
h
d
ialec
t
class
if
ica
tio
n
.
R
esear
ch
o
n
v
o
ice
r
ec
o
g
n
itio
n
tech
n
o
lo
g
y
r
elate
d
to
d
ialec
t
o
r
ac
ce
n
t
i
d
en
tific
atio
n
in
s
p
ec
if
ic
co
u
n
tr
ies
o
r
r
eg
io
n
s
h
as
b
ee
n
co
n
d
u
cted
ex
ten
s
iv
ely
.
E
x
am
p
les
in
clu
d
e
r
ec
o
g
n
izin
g
Ku
r
d
is
h
d
ialec
ts
u
s
in
g
1D
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN
)
[
9
]
an
d
id
en
tify
in
g
d
if
f
er
en
ce
s
b
etwe
en
two
C
o
lo
m
b
ian
d
ialec
ts
,
"
An
tio
q
u
eñ
o
"
an
d
"Bo
g
o
tan
o
,
"
u
s
in
g
C
NN
[
1
0
]
.
Similar
ly
,
in
I
n
d
o
n
esia
,
th
er
e
h
av
e
b
ee
n
s
tu
d
ies
f
o
cu
s
ed
o
n
d
ialec
t
o
r
r
eg
io
n
al
lan
g
u
a
g
e
class
if
icat
io
n
.
Fo
r
ex
am
p
le,
T
awa
q
al
an
d
Su
y
a
n
to
[
1
1
]
u
s
ed
a
d
ee
p
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
(
DR
NN)
to
id
en
tify
f
iv
e
m
ain
d
ialec
ts
:
J
av
an
ese
,
Su
n
d
an
ese
,
B
an
jar
,
B
u
g
in
ese,
an
d
Ma
lay
.
I
n
ad
d
itio
n
,
Nu
g
r
o
h
o
et
a
l
.
[
1
2
]
d
ev
elo
p
ed
a
d
ata
au
g
m
en
tatio
n
ap
p
r
o
ac
h
co
m
b
in
ed
with
a
s
ev
en
-
lay
e
r
d
ee
p
n
eu
r
al
n
etwo
r
k
(
DA
-
DNN7
L
)
to
class
if
y
eth
n
ic
s
p
ea
k
er
s
u
s
in
g
7
0
0
u
tter
a
n
ce
s
f
r
o
m
7
0
eth
n
ic
g
r
o
u
p
s
.
Oth
er
s
tu
d
ies
h
av
e
also
ad
d
r
ess
ed
d
i
alec
t
r
ec
o
g
n
itio
n
in
I
n
d
o
n
esia
n
lan
g
u
a
g
es,
s
u
ch
as
th
e
d
etec
tio
n
o
f
Su
n
d
a
n
ese
[
1
3
]
an
d
B
alin
ese
B
ad
u
n
g
[
1
4
]
.
Di
alec
t
id
en
tific
atio
n
in
v
o
lv
es
d
eter
m
i
n
in
g
t
h
e
d
iale
ct
ca
teg
o
r
y
o
f
s
p
o
k
en
u
tter
an
c
es.
T
h
is
task
f
o
cu
s
es
o
n
r
ec
o
g
n
izin
g
th
e
s
p
ea
k
er
’
s
r
eg
io
n
al
d
ialec
t w
ith
in
a
p
ar
tic
u
lar
lan
g
u
a
g
e
b
ased
s
o
lely
o
n
th
e
av
ailab
le
ac
o
u
s
tic
s
ig
n
als
[
1
5
]
.
T
h
is
s
tu
d
y
aim
s
to
id
en
tify
d
if
f
er
en
ce
s
am
o
n
g
v
ar
io
u
s
r
eg
io
n
al
d
ialec
ts
in
I
n
d
o
n
esia
th
r
o
u
g
h
v
o
ice
an
aly
s
is
.
A
cu
s
to
m
d
ataset
was
d
ev
elo
p
e
d
,
co
n
s
is
tin
g
o
f
s
ix
class
es
r
ep
r
esen
tin
g
d
ialec
ts
f
r
o
m
Me
d
an
,
Min
a
n
g
,
Su
n
d
a,
L
o
m
b
o
k
,
Ma
d
u
r
a,
a
n
d
Am
b
o
n
.
W
h
ile
th
e
d
ataset
p
r
o
v
id
es
a
f
o
u
n
d
atio
n
f
o
r
ex
p
l
o
r
i
n
g
th
ese
d
ialec
ts
,
it
r
ep
r
esen
ts
o
n
ly
a
s
u
b
s
et
o
f
I
n
d
o
n
esia
’
s
r
ich
lin
g
u
is
tic
d
iv
er
s
ity
,
war
r
an
tin
g
f
u
r
t
h
er
ex
p
an
s
io
n
in
f
u
tu
r
e
r
esear
ch
.
T
h
e
r
esear
ch
b
u
ild
s
u
p
o
n
p
r
i
o
r
wo
r
k
[
1
6
]
b
y
im
p
lem
e
n
tin
g
p
r
ev
io
u
s
ly
p
r
o
p
o
s
ed
tech
n
i
q
u
es,
in
clu
d
in
g
d
ata
au
g
m
en
tatio
n
m
et
h
o
d
s
(
s
u
ch
as
ad
d
in
g
n
o
is
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itch
s
h
if
tin
g
)
,
Me
l
-
f
r
eq
u
en
c
y
ce
p
s
tr
al
co
ef
f
icien
ts
(
MFC
C
)
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
,
an
d
c
o
m
p
ar
i
n
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
C
NN
an
d
m
u
ltil
ay
er
p
er
ce
p
t
r
o
n
(
ML
P)
m
o
d
els.
T
h
ese
tech
n
iq
u
es
wer
e
ap
p
lied
to
th
e
n
ewly
co
n
s
tr
u
cted
s
ix
-
class
d
at
aset
to
ev
alu
ate
th
e
co
n
s
is
ten
cy
o
f
th
eir
p
e
r
f
o
r
m
a
n
ce
o
n
n
ew
d
ata.
B
y
le
v
er
ag
i
n
g
th
ese
m
eth
o
d
s
,
th
e
s
tu
d
y
aim
ed
to
d
eter
m
i
n
e
wh
eth
er
th
e
alg
o
r
ith
m
s
co
u
ld
m
ain
tain
h
ig
h
ac
cu
r
ac
y
wh
en
ap
p
lied
to
a
d
if
f
e
r
en
t
d
ataset,
ef
f
ec
tiv
ely
en
a
b
lin
g
th
e
class
if
icatio
n
an
d
id
en
tific
atio
n
o
f
d
ialec
ts
b
ased
o
n
v
o
ic
e
f
ea
tu
r
es.
T
h
is
p
ap
er
is
s
tr
u
ctu
r
ed
as
f
o
llo
ws:
s
ec
tio
n
2
d
is
cu
s
s
e
s
th
e
s
t
ag
es
an
d
m
eth
o
d
s
ap
p
lied
in
t
h
is
s
tu
d
y
.
T
h
e
co
n
d
itio
n
s
a
n
d
r
esu
lts
o
f
th
e
ex
p
er
im
en
ts
ar
e
p
r
esen
te
d
i
n
s
ec
tio
n
3
.
Fin
ally
,
th
e
co
n
clu
s
io
n
o
f
th
is
r
esear
c
h
is
p
r
o
v
id
ed
i
n
s
ec
tio
n
4
.
2.
M
E
T
H
O
D
I
n
th
e
r
esear
ch
m
eth
o
d
o
lo
g
y
c
h
ap
ter
,
th
e
p
r
o
ce
s
s
o
r
s
cien
tifi
c
m
eth
o
d
u
s
ed
to
o
b
tain
d
ata
f
o
r
r
esear
c
h
p
u
r
p
o
s
es
is
d
ef
in
ed
.
T
h
is
m
eth
o
d
in
cl
u
d
es
s
cien
tific
ap
p
r
o
a
ch
es,
s
tep
s
,
an
d
ty
p
es,
as
well
as
th
e
lim
itatio
n
s
o
f
th
e
s
cien
tific
m
eth
o
d
.
Fig
u
r
e
1
illu
s
tr
a
tes th
e
s
tag
es o
f
th
e
r
esear
ch
th
at
will b
e
co
n
d
u
cte
d
i
n
th
is
s
tu
d
y
.
2
.
1
.
Da
t
a
s
et
A
d
ataset
is
a
co
llectio
n
o
f
d
at
a
th
at
p
r
o
v
i
d
es
an
o
v
er
v
iew
o
f
a
s
p
ec
if
ic
t
o
p
ic
[
1
7
]
.
T
h
e
d
ataset
u
s
ed
in
th
is
r
esear
ch
is
a
p
r
iv
ate
d
ataset,
n
am
ed
I
n
d
o
n
esia
n
d
ialec
ts
d
ataset
.
I
t
co
n
s
is
ts
o
f
d
ialec
ts
f
r
o
m
s
ev
er
al
r
e
g
io
n
al
lan
g
u
ag
es
in
I
n
d
o
n
esia
,
n
am
ely
Me
d
an
,
Min
an
g
,
Su
n
d
a,
L
o
m
b
o
k
,
Ma
d
u
r
a,
a
n
d
Am
b
o
n
.
T
h
e
d
ataset
co
n
tain
s
a
to
tal
o
f
1
,
9
9
6
f
iles
in
AA
C
f
o
r
m
at
[
1
7
]
.
T
h
is
d
ataset
was
co
llected
u
s
in
g
a
s
m
ar
tp
h
o
n
e
a
n
d
a
wir
eless
m
icr
o
p
h
o
n
e
o
v
er
a
p
er
io
d
o
f
2
m
o
n
th
s
,
a
n
d
th
e
f
iles
o
r
w
o
r
d
s
an
d
s
en
ten
ce
s
u
s
ed
ar
e
e
n
tire
ly
th
e
au
t
h
o
r
'
s
o
wn
,
wh
ich
th
e
au
th
o
r
co
m
p
iled
wit
h
in
1
wee
k
.
T
h
e
d
ataset
f
ea
tu
r
es
s
ix
s
p
ea
k
er
s
a
g
ed
b
e
twee
n
3
0
a
n
d
5
0
y
ea
r
s
o
ld
,
with
two
f
em
ale
an
d
f
o
u
r
m
ale
s
p
ea
k
er
s
[
1
8
]
.
T
h
e
s
p
ea
k
er
s
a
r
e
in
d
iv
id
u
als
wh
o
s
till
f
lu
en
tly
u
s
e
th
eir
r
eg
io
n
al
lan
g
u
ag
es,
co
m
p
lete
with
th
e
lo
ca
l
ac
ce
n
ts
.
E
ac
h
s
p
ea
k
er
will
d
eliv
er
an
av
er
ag
e
o
f
2
0
0
wo
r
d
s
o
r
s
en
ten
ce
s
th
at
h
av
e
b
ee
n
p
r
e
-
p
r
e
p
ar
ed
b
y
th
e
au
th
o
r
.
T
h
u
s
,
th
e
d
atab
a
s
e
co
n
tain
s
a
to
tal
o
f
1
,
9
9
6
u
t
ter
an
ce
s
,
with
ea
ch
d
ialec
t
av
er
ag
in
g
2
0
0
wo
r
d
s
/s
en
ten
ce
s
.
T
h
e
n
u
m
b
er
o
f
wo
r
d
s
co
n
s
is
ts
o
f
1
0
0
n
o
u
n
s
o
r
ad
jectiv
es,
5
0
in
ter
r
o
g
ativ
e
s
en
ten
c
es,
an
d
5
0
im
p
er
ativ
e
s
en
ten
ce
s
,
as selecte
d
f
o
r
t
h
is
d
ataset,
as sh
o
wn
in
T
ab
le
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
la
s
s
i
fica
tio
n
o
f reg
io
n
a
l la
n
g
u
a
g
e
d
i
a
lects u
s
in
g
co
n
v
o
lu
tio
n
a
l n
eu
r
a
l
n
etw
o
r
k
…
(
F
a
h
mi
B
.
Ma
r
a
s
a
b
ess
y)
5019
Fig
u
r
e
1
.
Stag
es o
f
re
s
ea
r
ch
T
ab
le
1
.
Vo
ca
b
u
lar
y
f
o
r
th
e
d
a
taset in
d
ialec
t la
n
g
u
ag
es
W
o
r
d
s
I
mp
e
r
a
t
i
v
e
s
e
n
t
e
n
c
e
s
I
n
t
e
r
r
o
g
a
t
i
v
e
s
e
n
t
e
n
c
e
s
F
a
t
h
e
r
P
l
e
a
s
e
d
o
n
't
b
e
n
o
i
s
y
,
k
i
d
s!
D
o
y
o
u
l
i
k
e
sp
i
c
y
f
o
o
d
?
S
i
b
l
i
n
g
N
o
sm
o
k
i
n
g
a
l
l
o
w
e
d
!
H
o
w
d
o
y
o
u
f
e
e
l
a
b
o
u
t
y
o
u
r
c
u
r
r
e
n
t
j
o
b
?
Li
o
n
P
l
e
a
s
e
m
a
k
e
b
l
a
c
k
c
o
f
f
e
e
f
o
r
f
a
t
h
e
r
!
W
h
e
n
a
r
e
y
o
u
p
l
a
n
n
i
n
g
t
o
l
e
a
v
e
?
R
i
v
e
r
D
o
n
o
t
s
p
e
a
k
w
i
t
h
o
u
t
p
e
r
m
i
ssi
o
n
i
n
t
h
e
me
e
t
i
n
g
r
o
o
m!
W
h
o
i
s
ma
k
i
n
g
n
o
i
se
f
r
o
m
t
h
e
n
e
x
t
r
o
o
m?
H
a
t
N
o
u
si
n
g
m
o
b
i
l
e
d
e
v
i
c
e
s w
h
i
l
e
d
r
i
v
i
n
g
!
W
h
y
d
o
y
o
u
a
l
w
a
y
s
g
e
t
a
n
g
r
y
e
v
e
r
y
t
i
me
I
mee
t
y
o
u
?
2
.
2
.
Da
t
a
prep
a
ra
t
i
o
n
Data
p
r
ep
ar
atio
n
in
v
o
l
v
es
co
n
v
er
tin
g
r
aw
d
ata
in
to
a
m
o
r
e
p
r
ac
tical
an
d
ef
f
icien
t f
o
r
m
,
en
s
u
r
in
g
th
at
th
e
d
ataset
to
b
e
p
r
o
ce
s
s
ed
an
d
an
aly
ze
d
is
b
o
th
ac
cu
r
ate
a
n
d
co
n
s
is
ten
t.
I
n
th
is
p
h
ase,
th
e
d
ialec
t
d
ataset
i
s
f
ir
s
t
d
iv
id
ed
,
an
d
th
e
n
d
ata
a
u
g
m
en
tatio
n
a
n
d
f
ea
tu
r
e
e
x
tr
ac
tio
n
ar
e
ap
p
lied
to
th
e
tr
ain
in
g
d
ata
to
m
ak
e
it
s
u
itab
le
f
o
r
th
e
p
r
o
p
o
s
ed
m
o
d
el.
Prio
r
to
f
ee
d
in
g
t
h
e
d
atase
t
in
to
th
e
m
o
d
el
alg
o
r
ith
m
,
d
a
ta
n
o
r
m
aliza
tio
n
is
ca
r
r
ied
o
u
t
u
s
in
g
s
tan
d
ar
d
s
ca
l
er
,
an
d
d
ialec
t
class
es
ar
e
en
c
o
d
ed
with
o
n
e
h
o
t
e
n
co
d
i
n
g
t
o
o
p
tim
ize
th
e
m
o
d
el'
s
d
ata
p
r
o
ce
s
s
in
g
.
I
n
t
h
e
in
itial
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e,
d
ata
s
p
li
ttin
g
is
p
er
f
o
r
m
e
d
t
o
d
iv
id
e
th
e
d
ataset
in
t
o
tw
o
p
ar
ts
:
tr
ain
in
g
d
ata
an
d
test
in
g
d
ata.
8
0
%
o
f
th
e
tr
ain
in
g
d
ata
will
b
e
u
s
ed
to
tr
ain
th
e
d
ev
el
o
p
ed
m
o
d
el,
wh
ile
th
e
r
e
m
ain
in
g
2
0
%
will
b
e
u
s
ed
as
v
alid
atio
n
an
d
test
in
g
d
ata
to
ev
alu
ate
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
b
ased
o
n
th
e
estab
lis
h
ed
p
er
f
o
r
m
an
ce
m
etr
ics.
T
h
e
r
an
d
o
m
s
am
p
lin
g
m
eth
o
d
will
b
e
ap
p
lied
to
en
s
u
r
e
th
at
th
e
d
ata
m
o
d
elin
g
p
r
o
ce
s
s
is
n
o
t b
iased
b
y
an
y
p
o
te
n
tial f
ea
tu
r
e
d
if
f
er
en
ce
s
in
th
e
d
ataset.
I
n
th
is
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e,
d
ata
au
g
m
en
tatio
n
will
b
e
ap
p
li
ed
to
th
e
tr
ain
in
g
d
ata
b
y
a
d
d
i
n
g
s
ev
er
al
d
ata
au
g
m
e
n
tatio
n
tech
n
iq
u
es
to
g
en
e
r
ate
n
ew
s
y
n
th
etic
d
at
a.
T
h
e
d
ata
au
g
m
en
tatio
n
m
eth
o
d
s
to
b
e
ap
p
lied
in
clu
d
e:
ad
d
in
g
n
o
is
e
with
a
n
o
is
e
r
atio
o
f
0
.
0
0
5
,
a
p
p
ly
in
g
tim
e
s
tr
etch
in
g
with
a
f
ac
to
r
o
f
0
.
8
,
an
d
s
h
if
tin
g
th
e
p
itch
with
n
_
s
tep
s
s
et
to
1
[
1
6
]
.
Af
ter
th
e
r
aw
d
ata
is
p
r
o
ce
s
s
e
d
th
r
o
u
g
h
au
g
m
en
tatio
n
,
th
e
n
ex
t
s
tep
is
to
ex
tr
ac
t
d
ata
b
y
ap
p
ly
i
n
g
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
o
n
th
e
MF
C
C
f
r
o
m
th
e
au
g
m
en
ted
d
ata
to
d
etec
t
lan
g
u
ag
e
d
ialec
ts
.
MFC
C
is
a
m
eth
o
d
u
s
ed
to
d
ec
o
m
p
o
s
e
s
p
ee
ch
s
ig
n
al
s
in
to
co
m
p
o
n
en
ts
to
r
ep
r
esen
t
in
f
o
r
m
ati
o
n
ab
o
u
t
p
itch
an
d
v
o
ca
l tr
ac
t c
h
a
r
ac
ter
is
tics
[
1
9
]
.
T
h
is
tech
n
iq
u
e
s
im
u
lates h
u
m
an
au
d
ito
r
y
b
eh
av
i
o
r
b
y
d
is
tin
g
u
is
h
in
g
s
o
u
n
d
f
r
eq
u
e
n
cies,
with
f
r
eq
u
en
cy
b
an
d
s
ca
lcu
lated
l
o
g
ar
ith
m
ically
[
2
0
]
.
T
h
e
p
r
o
ce
s
s
es
u
s
ed
in
f
ea
tu
r
e
ex
tr
ac
tio
n
at
th
e
MFC
C
lay
er
in
clu
d
e
pr
e
-
em
p
h
asize,
f
r
am
e
b
lo
ck
in
g
,
win
d
o
win
g
p
r
o
ce
s
s
,
f
ast
f
o
u
r
ier
tr
an
s
f
o
r
m
(
FF
T
)
,
Me
l
f
r
eq
u
e
n
cy
war
p
in
g
,
d
is
cr
ete
co
s
in
e
tr
a
n
s
f
o
r
m
(
DC
T
)
,
a
n
d
ce
p
s
tr
al
lif
ter
in
g
.
2
.
3
.
M
o
delin
g
T
h
e
ex
p
er
im
e
n
ts
in
v
o
lv
e
c
o
m
b
in
in
g
th
r
ee
d
ata
au
g
m
en
ta
tio
n
tech
n
iq
u
es
to
e
x
am
in
e
t
h
e
ef
f
ec
t
o
f
in
co
r
p
o
r
atin
g
ad
d
itio
n
al
d
ata
in
to
th
e
tr
ain
in
g
s
et
o
n
th
e
class
if
icatio
n
m
o
d
el'
s
p
er
f
o
r
m
an
c
e.
T
ab
le
2
o
u
tlin
es
th
e
ex
p
er
im
e
n
tal
m
o
d
els
to
b
e
co
n
d
u
cted
in
th
is
s
t
u
d
y
,
wh
ic
h
ar
e
b
ased
o
n
th
e
co
m
b
i
n
atio
n
o
f
a
d
d
in
g
n
o
is
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itch
s
h
if
tin
g
tech
n
iq
u
es
.
T
h
e
o
r
ig
in
al
d
ata
in
T
ab
le
2
r
ef
e
r
s
to
th
e
test
in
g
d
ata
th
at
h
as
b
ee
n
s
p
lit d
u
r
i
n
g
th
e
d
ata
p
r
e
p
ar
atio
n
s
tag
e.
E
x
p
er
im
en
ts
will
b
e
c
o
n
d
u
ct
ed
o
n
eig
h
t
m
o
d
els,
ea
c
h
o
f
wh
ich
im
p
lem
en
ts
th
e
MF
C
C
f
ea
tu
r
e
ex
tr
ac
tio
n
tec
h
n
iq
u
e
.
T
h
ese
m
o
d
els
ar
e
d
esig
n
ed
to
e
v
al
u
ate
th
e
im
p
ac
t
o
f
d
if
f
er
en
t
d
ata
a
u
g
m
en
tatio
n
co
m
b
in
atio
n
s
o
n
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
.
T
h
is
s
tu
d
y
will
co
n
d
u
ct
d
ialec
t
class
if
icatio
n
m
o
d
elin
g
u
s
in
g
th
e
C
NN
an
d
ML
P a
lg
o
r
ith
m
s
ad
o
p
ted
f
r
o
m
p
r
ev
io
u
s
s
tu
d
ies
[
1
6
]
.
T
h
e
ar
c
h
itectu
r
e
o
f
th
e
ML
P
m
o
d
el
will
b
e
c
o
m
p
ar
e
d
with
th
e
p
r
o
p
o
s
ed
C
NN
m
o
d
el
a
r
ch
itectu
r
e.
ML
P
is
k
n
o
wn
f
o
r
its
s
tr
o
n
g
s
ca
lab
ilit
y
an
d
ef
f
icien
cy
in
lear
n
in
g
p
atter
n
s
co
m
p
ar
ed
to
o
t
h
e
r
class
if
ier
s
,
th
an
k
s
to
its
co
m
p
ac
t stru
ctu
r
e
an
d
ad
ap
tiv
e
m
ec
h
a
n
is
m
s
[
2
1
]
.
T
h
e
ML
P m
o
d
el
u
s
ed
h
as two
h
id
d
en
lay
er
s
,
with
4
8
8
n
o
d
es
in
th
e
f
ir
s
t
lay
er
a
n
d
4
4
3
n
o
d
es
in
t
h
e
s
ec
o
n
d
.
T
h
e
ML
P
ar
ch
itectu
r
e
will
b
e
a
p
p
lied
to
t
h
e
eig
h
t
ex
p
er
im
en
tal
m
o
d
els
with
a
r
an
d
o
m
s
tate
p
ar
am
eter
s
et
to
1
,
a
m
ax
im
u
m
n
u
m
b
er
o
f
iter
atio
n
s
o
f
3
2
5
,
a
lear
n
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
1
7
-
5
0
2
6
5020
r
ate
o
f
0
.
0
0
2
3
3
,
n
_
iter
_
n
o
_
ch
an
g
e
=
2
7
,
an
d
an
alp
h
a
v
alu
e
o
f
0
.
0
0
1
8
5
f
o
r
L
2
r
e
g
u
lar
izati
o
n
.
T
h
e
v
ar
iatio
n
o
f
th
e
ML
P
ar
ch
itectu
r
e
with
p
a
r
am
eter
s
an
d
t
h
eir
v
al
u
es
is
an
ar
ch
itectu
r
e
th
at
h
as
b
ee
n
ca
r
r
ied
o
u
t
b
y
p
r
e
v
io
u
s
s
tu
d
ies
[
2
2
]
.
T
h
e
ar
ch
itectu
r
e
o
f
th
is
m
o
d
e
l
was
d
ev
elo
p
ed
u
s
in
g
a
C
N
N
im
p
lem
en
ted
with
th
e
Ker
as
lib
r
ar
y
in
Py
th
o
n
.
C
NN
is
a
v
ar
iatio
n
o
f
th
e
MLP
in
s
p
ir
ed
b
y
th
e
h
u
m
a
n
n
eu
r
al
n
etwo
r
k
[
2
3
]
.
I
n
th
is
s
tu
d
y
,
th
e
p
r
o
p
o
s
ed
ar
ch
itectu
r
e
em
p
l
o
y
s
a
n
1
D
C
NN
.
T
h
e
co
n
f
ig
u
r
atio
n
o
f
th
e
m
o
d
el
co
n
s
is
ts
o
f
4
0
in
p
u
t
n
e
u
r
o
n
s
,
f
o
u
r
h
id
d
e
n
lay
er
s
with
3
2
,
6
4
,
6
4
,
an
d
6
4
h
id
d
en
n
eu
r
o
n
s
r
esp
ec
tiv
ely
,
an
d
is
tr
ain
ed
f
o
r
3
0
0
ep
o
c
h
s
with
a
b
atch
s
ize
o
f
1
6
.
T
h
e
r
e
c
t
i
f
i
e
d
l
i
n
e
a
r
u
n
i
t
(
R
e
L
U
)
ac
tiv
atio
n
f
u
n
ctio
n
is
ap
p
lied
,
alo
n
g
with
d
r
o
p
o
u
t
r
ates o
f
0
.
0
7
an
d
0
.
1
4
t
o
p
r
ev
en
t
o
v
er
f
itti
n
g
.
T
h
e
m
o
d
el
o
p
tim
izatio
n
u
s
es th
e
Ad
am
o
p
tim
izer
with
a
lear
n
in
g
r
ate
o
f
0
.
0
0
0
1
[
1
6
]
.
T
h
e
n
ex
t
s
tep
is
to
b
u
ild
th
e
C
NN
ar
ch
itectu
r
e
b
ased
o
n
t
h
e
m
o
d
el
p
a
r
am
eter
s
th
at
h
a
v
e
b
e
en
s
elec
ted
an
d
p
r
o
p
o
s
ed
f
o
r
th
e
class
if
ic
atio
n
m
o
d
el.
T
h
e
co
n
s
tr
u
ctio
n
o
f
th
e
C
NN
ar
ch
itectu
r
e
is
p
r
esen
ted
in
Fig
u
r
e
2
.
T
h
is
f
ig
u
r
e
p
r
o
v
i
d
es a
v
is
u
al
r
ep
r
esen
tatio
n
o
f
th
e
lay
er
s
an
d
co
n
f
ig
u
r
atio
n
s
ap
p
lied
in
th
e
m
o
d
el.
T
ab
le
2
.
E
x
p
er
im
en
tal
se
tu
p
[
1
6
]
M
o
d
e
l
D
a
t
a
a
u
g
me
n
t
a
t
i
o
n
F
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
A
Tr
a
i
n
i
n
g
d
a
t
a
M
F
C
C
B
Tr
a
i
n
i
n
g
d
a
t
a
+
a
d
d
i
n
g
n
o
i
se
M
F
C
C
C
Tr
a
i
n
i
n
g
d
a
t
a
+
t
i
m
e
s
t
r
e
t
c
h
i
n
g
M
F
C
C
D
Tr
a
i
n
i
n
g
d
a
t
a
+
p
i
t
c
h
s
h
i
f
t
i
n
g
M
F
C
C
E
Tr
a
i
n
i
n
g
d
a
t
a
+
a
d
d
i
n
g
n
o
i
se
+
t
i
me
st
r
e
t
c
h
i
n
g
M
F
C
C
F
Tr
a
i
n
i
n
g
d
a
t
a
+
a
d
d
i
n
g
n
o
i
s
e
+
p
i
t
c
h
s
h
i
f
t
i
n
g
M
F
C
C
G
Tr
a
i
n
i
n
g
d
a
t
a
+
t
i
m
e
s
t
r
e
t
c
h
i
n
g
+
p
i
t
c
h
s
h
i
f
t
i
n
g
M
F
C
C
H
Tr
a
i
n
i
n
g
d
a
t
a
+
a
d
d
i
n
g
n
o
i
se
+
t
i
m
e
s
t
r
e
t
c
h
i
n
g
+
p
i
t
c
h
s
h
i
f
t
i
n
g
M
F
C
C
Fig
u
r
e
2
.
C
o
n
v
o
lu
tio
n
al
n
eu
r
a
l n
etwo
r
k
ar
c
h
itectu
r
e
T
h
is
s
tu
d
y
ad
o
p
ts
an
ar
c
h
itectu
r
e
b
ased
o
n
p
r
ev
io
u
s
r
esear
ch
[
1
6
]
,
as
illu
s
tr
ated
in
Fig
u
r
e
2
.
T
h
e
d
esig
n
f
ea
tu
r
es
a
o
n
e
-
d
im
e
n
s
io
n
al
C
NN
f
o
r
class
if
y
in
g
h
u
m
an
d
ialec
ts
u
s
in
g
f
ea
tu
r
es
ex
tr
ac
ted
f
r
o
m
au
d
io
f
iles
.
T
h
is
1
D
C
NN
d
e
s
ig
n
ac
ce
p
ts
in
p
u
t
d
ata
in
th
e
f
o
r
m
o
f
a
4
0
×1
ar
r
ay
,
co
r
r
esp
o
n
d
in
g
to
f
ea
tu
r
es
ex
tr
ac
ted
f
r
o
m
th
e
d
ialec
t
d
ataset
.
T
h
e
m
o
d
el
c
o
n
s
is
t
s
o
f
f
o
u
r
h
i
d
d
en
lay
er
s
,
ea
c
h
with
a
k
e
r
n
el
s
ize
o
f
3
,
co
n
f
i
g
u
r
ed
with
3
2
,
6
4
,
6
4
,
an
d
6
4
n
eu
r
o
n
s
,
r
esp
ec
tiv
e
ly
.
T
h
e
ac
tiv
atio
n
f
u
n
ctio
n
f
o
r
th
e
in
p
u
t
a
n
d
h
id
d
en
lay
er
s
is
R
eL
U.
T
o
ad
d
r
ess
o
v
er
f
itti
n
g
,
th
e
m
o
d
el
in
co
r
p
o
r
ates
b
atch
n
o
r
m
aliza
tio
n
,
L
2
r
eg
u
lar
izatio
n
(
v
alu
e:
0
.
0
0
0
1
)
,
an
d
d
r
o
p
o
u
t
r
ates
o
f
0
.
0
7
a
n
d
0
.
1
4
ap
p
lied
to
s
p
ec
if
ic
lay
er
s
.
T
h
e
ar
ch
itectu
r
e
also
in
clu
d
es a
m
a
x
p
o
o
lin
g
la
y
er
with
a
p
o
o
l size
o
f
2
an
d
a
f
u
lly
co
n
n
ec
ted
o
u
t
p
u
t
lay
er
with
s
ix
u
n
its
,
co
r
r
esp
o
n
d
in
g
to
th
e
s
ix
d
ialec
t
clas
s
es.
T
h
e
o
u
tp
u
t
lay
er
u
s
es
th
e
So
f
tMa
x
ac
tiv
atio
n
f
u
n
ctio
n
f
o
r
class
if
icatio
n
.
Op
t
im
izatio
n
is
ca
r
r
ied
o
u
t
u
s
in
g
th
e
Ad
am
alg
o
r
ith
m
with
a
lear
n
in
g
r
ate
o
f
0
.
0
0
0
1
,
an
d
ca
teg
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
is
u
s
ed
as
th
e
lo
s
s
f
u
n
ctio
n
.
T
h
e
m
o
d
el
is
tr
ain
e
d
f
o
r
3
0
0
ep
o
c
h
s
with
a
b
atch
s
i
ze
o
f
1
6
,
an
d
a
Mo
d
elC
h
ec
k
p
o
in
t
ca
llb
ac
k
is
u
s
ed
to
s
av
e
t
h
e
b
est
-
p
er
f
o
r
m
in
g
m
o
d
el
b
ased
o
n
th
e
l
o
west v
alid
atio
n
lo
s
s
d
u
r
in
g
tr
ain
in
g
.
2.
4
.
E
v
a
lua
t
i
o
n
Af
ter
co
n
d
u
ctin
g
th
e
ex
p
e
r
im
en
ts
,
th
e
n
ex
t
s
tep
is
to
ev
alu
ate
ea
ch
e
x
p
er
im
e
n
tal
m
o
d
el
u
s
in
g
C
NN
an
d
ML
P.
T
h
e
ev
alu
atio
n
w
ill
b
e
p
er
f
o
r
m
ed
u
s
in
g
test
in
g
d
ata.
T
h
e
ev
alu
atio
n
p
r
o
ce
s
s
wi
ll
u
tili
ze
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
la
s
s
i
fica
tio
n
o
f reg
io
n
a
l la
n
g
u
a
g
e
d
i
a
lects u
s
in
g
co
n
v
o
lu
tio
n
a
l n
eu
r
a
l
n
etw
o
r
k
…
(
F
a
h
mi
B
.
Ma
r
a
s
a
b
ess
y)
5021
Scik
it
-
lear
n
lib
r
ar
ies.
T
h
e
p
er
f
o
r
m
an
ce
m
etr
ics
u
s
ed
f
o
r
m
o
d
el
ev
alu
atio
n
ar
e
ac
c
u
r
ac
y
,
F1
-
s
co
r
e,
p
r
ec
is
io
n
,
r
ec
all,
an
d
k
a
p
p
a.
T
h
e
s
elec
tio
n
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
m
o
d
el
will
b
e
d
eter
m
in
e
d
b
ased
o
n
th
e
p
e
r
f
o
r
m
an
c
e
m
etr
ics
,
with
th
e
b
est
r
esu
lts
f
r
o
m
th
e
ei
g
h
t
ex
p
er
im
en
tal
m
o
d
els
in
T
ab
le
2
.
Fu
r
th
er
m
o
r
e,
in
th
e
C
NN
alg
o
r
ith
m
,
to
en
s
u
r
e
th
at
th
e
s
elec
ted
m
o
d
el
is
n
o
t
o
v
er
f
itti
n
g
,
th
e
ex
p
er
im
en
t
p
r
o
c
ess
in
clu
d
es
p
lo
ttin
g
tr
ain
in
g
an
d
test
in
g
b
ased
o
n
th
e
h
is
to
r
y
o
f
ac
c
u
r
ac
y
a
n
d
lo
s
s
.
T
h
e
f
o
r
m
u
la
f
o
r
f
in
d
in
g
t
h
e
m
ea
s
u
r
em
en
t
m
etr
ic
v
alu
e
u
s
ed
in
m
u
lticlas
s
class
if
icatio
n
is
as f
o
llo
ws
[
1
6
]
:
A
c
c
ura
c
y
=
∑
(
TP
k
+
TN
k
)
TP
k
+
FP
k
+
TN
k
+
FN
k
K
k
=
1
(
1
)
In
(
1
)
,
it
m
ea
s
u
r
es
th
e
p
r
o
p
o
r
tio
n
o
f
o
v
e
r
all
co
r
r
ec
t
p
r
e
d
ict
io
n
s
to
th
e
to
tal
p
r
ed
ictio
n
s
.
K
is
th
e
n
u
m
b
er
o
f
class
es
;
T
Pₖ
i
s
th
e
tr
u
e
p
o
s
itiv
e
f
o
r
class
k
;
T
Nₖ
is
th
e
tr
u
e
n
eg
ativ
e;
FP
ₖ
is
th
e
f
alse
p
o
s
itiv
e;
an
d
FNₖ
is
th
e
f
alse n
eg
ati
v
e.
T
o
ev
alu
ate
m
o
d
el
p
er
f
o
r
m
a
n
ce
m
o
r
e
co
m
p
r
eh
e
n
s
iv
ely
ac
r
o
s
s
all
clas
s
es,
s
ev
er
al
ad
d
itio
n
al
m
etr
ics
ar
e
u
s
ed
.
In
(
2
)
d
ef
in
es
th
e
m
a
cr
o
av
e
r
ag
e
p
r
ec
is
io
n
,
wh
ic
h
c
alcu
lates
th
e
av
er
ag
e
p
r
ec
is
io
n
ac
r
o
s
s
all
class
es
b
y
tak
in
g
th
e
m
ea
n
o
f
in
d
iv
i
d
u
al
class
p
r
ec
is
io
n
s
co
r
es.
In
(
3
)
r
ep
r
esen
ts
th
e
m
ac
r
o
av
er
ag
e
r
ec
all,
wh
ich
av
er
ag
es th
e
r
ec
all
v
alu
es f
o
r
ea
ch
class
,
r
ef
lectin
g
th
e
m
o
d
el’
s
ab
ilit
y
to
co
r
r
ec
tly
i
d
en
tif
y
p
o
s
itiv
e
in
s
tan
ce
s
.
M
a
c
r
o
A
ve
r
a
ge
Pr
e
c
ision
=
∑
TP
k
TP
k
+
FP
k
K
k
=
1
K
(
2
)
M
a
c
r
o
A
ve
r
a
ge
R
e
c
a
l
l
=
∑
TP
k
TP
k
+
FN
k
K
k
=
1
K
(
3
)
T
h
e
b
alan
ce
b
etwe
en
t
h
ese
two
m
etr
ics
is
ca
p
tu
r
e
d
th
r
o
u
g
h
t
h
e
m
ac
r
o
F1
-
s
co
r
e
as
s
h
o
wn
i
n
(
4
)
,
wh
ich
p
r
o
v
id
es
a
h
ar
m
o
n
ic
m
ea
n
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
al
l
to
en
s
u
r
e
f
air
n
ess
in
m
u
lt
i
-
class
ev
alu
atio
n
.
Fu
r
th
er
m
o
r
e
,
in
(
5
)
p
r
esen
ts
C
o
h
en
’
s
Kap
p
a
(
κ)
m
ea
s
u
r
es
th
e
ag
r
ee
m
e
n
t
b
etwe
en
th
e
m
o
d
el
p
r
e
d
ictio
n
s
an
d
th
e
ac
tu
al
lab
els
wh
ile
co
n
s
id
er
in
g
th
e
p
o
s
s
ib
ilit
y
o
f
ag
r
ee
m
e
n
t
o
cc
u
r
r
i
n
g
b
y
ch
an
ce
.
c
is
th
e
n
u
m
b
er
o
f
co
r
r
ec
t
p
r
ed
ictio
n
s
(
d
iag
o
n
al
elem
en
t
s
o
f
th
e
co
n
f
u
s
io
n
m
atr
ix
)
,
s
is
th
e
to
tal
n
u
m
b
er
o
f
s
am
p
les,
p
ₖ
i
s
th
e
n
u
m
b
er
o
f
p
r
ed
ictio
n
s
f
o
r
class
k
,
an
d
tₖ
is
th
e
ac
tu
al
n
u
m
b
e
r
o
f
s
am
p
le
s
f
o
r
class
k
.
M
a
c
r
o
F1
−
s
c
or
e
=
2
(
Mac
r
o
A
v
er
ag
e
Pr
ecis
i
o
n
×
Mac
r
o
A
v
er
ag
e
Recal
l
Mac
r
o
A
v
er
a
g
e
Pr
ecis
i
o
n
+
Mac
r
o
A
v
er
ag
e
Recal
l
)
(
4
)
C
ohe
n
′
s
Ka
pp
a
=
c
×
s
−
∑
p
k
×
t
k
K
k
s
2
−
∑
p
k
×
t
k
K
k
(
5
)
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
s
y
s
tem
co
m
p
o
n
en
ts
u
s
e
d
f
o
r
th
e
ex
p
er
im
en
ts
in
th
is
r
esear
ch
ar
e
s
p
ec
if
ied
as
f
o
llo
ws:
t
h
e
o
p
er
atin
g
s
y
s
tem
is
W
in
d
o
ws
1
0
Ho
m
e
Sin
g
le
L
an
g
u
ag
e,
r
u
n
n
in
g
o
n
an
I
n
tel(
R
)
C
o
r
e(
T
M)
i3
-
3
1
2
0
M
C
PU
@
2
.
5
0
GHz
p
r
o
ce
s
s
o
r
.
T
h
e
s
y
s
tem
is
eq
u
ip
p
ed
with
6
GB
o
f
R
AM
an
d
a
2
9
8
GB
h
ar
d
d
is
k
.
Fo
r
d
ata
m
i
n
in
g
ap
p
licatio
n
s
,
Go
o
g
le
C
o
lab
was u
tili
ze
d
to
s
u
p
p
o
r
t t
h
e
ex
p
e
r
im
en
tatio
n
an
d
an
aly
s
is
p
r
o
ce
s
s
.
3
.
1
.
Da
t
a
s
et
ex
p
lo
r
a
t
io
n
T
h
e
f
ir
s
t
s
tep
co
n
d
u
cted
i
n
th
e
ex
p
er
im
e
n
t
o
f
t
h
is
r
esear
ch
w
as
p
er
f
o
r
m
i
n
g
a
s
tatis
tical
an
aly
s
is
o
f
th
e
d
ataset
th
r
o
u
g
h
th
e
ex
p
l
o
r
ato
r
y
d
ata
an
aly
s
is
(
E
DA)
p
r
o
ce
s
s
.
T
h
e
d
ataset
co
n
s
is
ts
o
f
s
i
x
d
ialec
t
lan
g
u
ag
e
class
es
:
Am
b
o
n
,
L
o
m
b
o
k
,
M
ad
u
r
a,
Me
d
an
,
Min
an
g
,
an
d
S
u
n
d
a.
Af
ter
ex
p
lo
r
in
g
th
e
d
at
a,
th
e
Am
b
o
n
class
co
n
tain
s
2
0
0
au
d
io
d
ata,
th
e
L
o
m
b
o
k
class
h
as
2
0
0
au
d
i
o
d
a
ta,
an
d
th
e
Ma
d
u
r
a
class
co
n
tain
s
1
9
7
au
d
io
d
ata
d
u
e
to
th
e
v
o
ice
r
ec
o
r
d
in
g
p
r
o
ce
s
s
.
T
h
e
Me
d
an
class
h
as
2
0
0
au
d
i
o
d
ata,
th
e
Min
a
n
g
class
co
n
ta
in
s
1
9
9
a
u
d
io
d
ata,
an
d
th
e
Su
n
d
a
class
h
as
2
0
0
au
d
i
o
d
ata.
T
h
e
to
tal
n
u
m
b
er
o
f
d
ata
u
s
ed
is
1
,
9
9
6
,
all
o
f
wh
ich
ar
e
au
d
i
o
d
ata
in
AAC
f
o
r
m
at.
Ho
wev
e
r
,
th
e
to
tal
n
u
m
b
er
o
f
d
atasets
,
wh
ich
s
h
o
u
ld
h
a
v
e
b
ee
n
1
,
2
0
0
,
d
ec
r
ea
s
ed
b
ec
au
s
e
s
o
m
e
d
ata
wer
e
lo
s
t d
u
r
in
g
th
e
tr
an
s
f
er
p
r
o
ce
s
s
f
r
o
m
th
e
s
m
ar
tp
h
o
n
e
to
th
e
lap
t
o
p
/co
m
p
u
te
r
.
3
.
2
.
Resul
t
o
f
da
t
a
prepa
ra
t
i
o
n
B
ef
o
r
e
co
n
d
u
ctin
g
ex
p
er
im
en
ts
f
o
r
m
o
d
elin
g
,
th
e
d
ataset
u
n
d
er
g
o
es
a
d
ata
p
r
ep
ar
atio
n
s
tag
e.
T
h
is
p
h
ase
is
ess
en
tial
to
en
s
u
r
e
t
h
at
th
e
d
ata
is
r
ea
d
y
f
o
r
p
r
o
ce
s
s
in
g
b
y
th
e
class
if
icatio
n
m
o
d
els.
I
t
in
cl
u
d
es
p
r
o
ce
d
u
r
es su
ch
as sp
litt
in
g
th
e
d
ataset
an
d
ap
p
l
y
in
g
a
u
g
m
e
n
tatio
n
tech
n
iq
u
es.
T
h
e
d
ataset
h
as
b
ee
n
d
iv
id
e
d
in
to
tr
ain
in
g
a
n
d
test
in
g
d
ata,
wh
er
e
th
e
tr
ain
in
g
d
ata
is
u
s
ed
to
tr
ain
th
e
m
o
d
el,
an
d
th
e
test
in
g
d
ata
is
u
s
ed
to
ev
alu
ate
its
p
er
f
o
r
m
a
n
c
e.
Ou
t
o
f
th
e
to
tal
1
,
9
9
6
au
d
i
o
d
ata
in
th
e
l
a
n
g
u
a
g
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
1
7
-
5
0
2
6
5022
d
ialec
t
d
ataset
,
8
0
% o
r
9
5
6
au
d
io
d
ata
wer
e
allo
ca
te
d
f
o
r
tr
a
in
in
g
,
wh
ile
2
0
% o
r
2
4
0
au
d
io
d
ata
wer
e
u
s
ed
f
o
r
test
in
g
.
T
h
e
d
is
tr
ib
u
tio
n
o
f
d
at
a
ac
r
o
s
s
th
e
s
ix
d
ialec
t
cla
s
s
e
s
is
r
elativ
ely
b
alan
ce
d
.
Fo
r
th
e
tr
ain
in
g
s
et,
Am
b
o
n
co
n
tr
ib
u
ted
1
5
9
s
am
p
les,
L
o
m
b
o
k
1
6
5
s
am
p
les,
Ma
d
u
r
a
1
6
2
s
am
p
les,
Me
d
an
1
5
0
s
a
m
p
les,
Min
an
g
1
6
1
s
am
p
les,
an
d
Su
n
d
a
1
5
9
s
am
p
les.
Me
an
wh
ile,
f
o
r
th
e
te
s
tin
g
s
et,
Am
b
o
n
c
o
n
s
is
ted
o
f
4
1
s
a
m
p
les,
L
o
m
b
o
k
3
5
s
am
p
les,
Ma
d
u
r
a
3
5
s
am
p
les,
Me
d
an
5
0
s
am
p
les,
Min
an
g
3
8
s
am
p
les,
an
d
Su
n
d
a
4
1
s
am
p
l
es
.
T
h
is
d
is
tr
ib
u
tio
n
en
s
u
r
es th
at
ea
ch
d
ialec
t is f
ai
r
ly
r
ep
r
esen
te
d
in
b
o
th
th
e
tr
ai
n
in
g
an
d
test
in
g
p
h
ases
.
Data
au
g
m
en
tatio
n
is
a
tech
n
iq
u
e
u
s
ed
to
g
en
er
ate
n
ew
d
ata
f
r
o
m
ex
is
tin
g
s
am
p
les.
I
n
th
is
s
tu
d
y
,
t
h
r
ee
au
g
m
en
tatio
n
m
eth
o
d
s
wer
e
ap
p
lied
to
en
h
an
ce
th
e
tr
ai
n
in
g
s
et:
ad
d
in
g
n
o
is
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itc
h
s
h
if
tin
g
.
T
h
ese
m
eth
o
d
s
h
elp
i
n
cr
ea
s
e
th
e
d
iv
er
s
i
ty
o
f
tr
ain
in
g
d
ata
an
d
im
p
r
o
v
e
m
o
d
el
g
e
n
er
aliza
tio
n
.
Fig
u
r
e
3
p
r
esen
ts
a
co
m
p
ar
a
tiv
e
v
is
u
aliza
tio
n
o
f
wav
ef
o
r
m
r
ep
r
esen
tatio
n
s
b
ef
o
r
e
an
d
af
ter
th
e
ap
p
licatio
n
o
f
d
ata
a
u
g
m
e
n
tatio
n
tech
n
i
q
u
es to
th
e
au
d
i
o
s
ig
n
als.
I
n
p
a
r
ticu
lar
,
Fig
u
r
e
3
(
a
)
d
ep
icts
th
e
o
r
ig
i
n
al
wav
ef
o
r
m
co
r
r
esp
o
n
d
in
g
to
an
Am
b
o
n
d
ialec
t
s
am
p
le,
s
er
v
in
g
as
th
e
b
aselin
e
r
ef
e
r
en
ce
.
Fi
g
u
r
e
3
(
b
)
illu
s
tr
ates
th
e
wav
ef
o
r
m
f
o
llo
win
g
th
e
in
tr
o
d
u
ctio
n
o
f
a
d
d
itiv
e
Ga
u
s
s
ian
n
o
is
e
with
an
am
p
litu
d
e
p
ar
am
eter
o
f
0
.
0
0
5
,
wh
ich
v
is
ib
ly
in
cr
ea
s
es
s
ig
n
al
v
ar
iab
ilit
y
.
Fig
u
r
e
3
(
c)
d
em
o
n
s
tr
ates
th
e
im
p
ac
t
o
f
tim
e
s
tr
et
ch
i
n
g
,
ev
id
e
n
ce
d
b
y
th
e
elo
n
g
atio
n
o
f
th
e
wav
e
f
o
r
m
d
u
r
atio
n
to
ap
p
r
o
x
im
ately
2
.
8
s
ec
o
n
d
s
wh
ile
p
r
eser
v
in
g
tem
p
o
r
al
p
atter
n
s
.
Fig
u
r
e
3
(
d
)
d
is
p
la
y
s
t
h
e
wa
v
e
f
o
r
m
af
te
r
ap
p
l
y
i
n
g
p
itc
h
s
h
if
tin
g
,
w
h
i
c
h
alt
er
s
t
h
e
f
r
eq
u
e
n
cy
c
o
n
te
n
t
wi
th
o
u
t
ch
an
g
i
n
g
t
h
e
t
em
p
o
r
al
s
ca
le
.
T
h
e
r
ed
a
n
n
o
tat
io
n
s
in
f
ig
u
r
e
h
i
g
h
li
g
h
t
s
p
e
ci
f
ic
r
e
g
i
o
n
s
w
h
e
r
e
n
o
ta
b
l
e
d
i
f
f
er
e
n
ce
s
in
a
m
p
lit
u
d
e
a
n
d
f
r
e
q
u
e
n
c
y
c
h
ar
ac
t
er
is
tics
a
r
e
o
b
s
e
r
v
a
b
le
b
et
wee
n
th
e
o
r
i
g
i
n
a
l
an
d
a
u
g
m
en
t
ed
s
i
g
n
als
.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
3
.
W
av
ep
lo
t
v
is
u
aliza
tio
n
b
ef
o
r
e
a
n
d
af
ter
ap
p
ly
i
n
g
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es: (
a)
o
r
ig
in
al
wav
ep
lo
t sam
p
le
f
r
o
m
th
e
Am
b
o
n
d
ialec
t,
(
b
)
wa
v
ep
lo
t a
f
ter
ad
d
in
g
n
o
is
e,
(
c)
wav
ep
lo
t a
f
ter
ap
p
ly
in
g
tim
e
s
tr
etch
in
g
,
an
d
(
d
)
wav
ep
lo
t a
f
ter
ap
p
ly
in
g
p
itch
s
h
if
tin
g
Ap
p
ly
in
g
th
e
th
r
ee
d
ata
au
g
m
e
n
tatio
n
tech
n
i
q
u
es
to
9
5
6
au
d
io
s
am
p
les
f
o
r
m
o
d
el
tr
ain
in
g
p
r
o
d
u
ce
d
a
to
tal
o
f
s
ev
en
ex
p
er
im
e
n
tal
m
o
d
els,
in
ad
d
itio
n
to
o
n
e
m
o
d
el
th
at
u
tili
ze
d
o
n
ly
th
e
o
r
ig
in
al
tr
ain
in
g
d
ata.
E
x
p
er
im
en
tal
m
o
d
el
A
u
s
es
th
e
9
5
6
o
r
ig
i
n
al
au
d
io
d
ata
with
o
u
t
au
g
m
e
n
tatio
n
.
Mo
d
el
B
ap
p
lies
th
e
ad
d
n
o
is
e
tech
n
iq
u
e,
in
c
r
ea
s
in
g
th
e
d
ataset
to
1
,
9
1
2
s
am
p
les.
Mo
d
el
C
ap
p
lies
th
e
tim
e
s
tr
etch
tech
n
iq
u
e,
also
p
r
o
d
u
cin
g
1
,
9
1
2
s
am
p
les,
wh
ile
m
o
d
el
D
ap
p
lies
th
e
p
itch
s
h
if
t
tech
n
iq
u
e,
r
esu
ltin
g
in
t
h
e
s
am
e
n
u
m
b
er
o
f
s
am
p
les.
Mo
d
el
E
co
m
b
in
es
ad
d
n
o
is
e
an
d
tim
e
s
tr
etch
tech
n
iq
u
es
,
e
x
p
an
d
i
n
g
th
e
d
ataset
to
2
,
8
6
8
s
am
p
les.
Similar
ly
,
m
o
d
el
F
ap
p
lies
ad
d
n
o
is
e
wi
th
p
itch
s
h
if
t
,
an
d
m
o
d
el
G
ap
p
lies
tim
e
s
tr
etch
wi
th
p
itch
s
h
if
t
,
b
o
th
y
ield
in
g
2
,
8
6
8
s
am
p
les.
Fin
ally
,
m
o
d
e
l
H
in
teg
r
ates
all
th
r
ee
tech
n
iq
u
es
(
ad
d
n
o
is
e,
tim
e
s
tr
etch
,
an
d
p
itch
s
h
if
t
)
,
p
r
o
d
u
cin
g
th
e
lar
g
est d
ataset
with
3
,
8
2
4
au
d
io
s
am
p
les.
Featu
r
e
ex
tr
ac
tio
n
was
p
er
f
o
r
m
ed
af
ter
th
e
d
ata
au
g
m
en
t
atio
n
p
r
o
ce
s
s
.
T
o
g
e
n
er
ate
MFC
C
,
th
e
L
ib
r
o
s
a
lib
r
ar
y
in
Py
t
h
o
n
was
u
s
ed
.
MFC
C
is
d
esig
n
ed
to
r
ef
lect
h
u
m
an
p
er
ce
p
tio
n
o
f
f
r
e
q
u
en
cy
b
y
co
n
v
er
tin
g
co
n
v
en
tio
n
al
f
r
eq
u
e
n
cy
t
o
th
e
Me
l
s
ca
le.
T
h
e
p
r
o
ce
s
s
b
e
g
in
s
with
a
p
r
e
-
e
m
p
h
asis
s
tag
e
to
p
r
eser
v
e
h
ig
h
f
r
eq
u
e
n
cies
th
at
ar
e
u
s
u
ally
lo
s
t
d
u
r
in
g
s
o
u
n
d
p
r
o
d
u
ctio
n
.
T
h
e
au
d
io
s
ig
n
al
is
th
en
d
i
v
id
ed
i
n
to
4
0
f
r
am
es.
E
ac
h
f
r
am
e
is
weig
h
ted
u
s
in
g
a
win
d
o
w
f
u
n
ctio
n
,
with
th
e
au
d
i
o
s
ig
n
al
s
p
lit
in
to
o
v
er
lap
p
i
n
g
wi
n
d
o
ws
b
ased
o
n
th
e
p
ar
am
eter
s
n
_
f
f
t
o
f
2
0
4
8
an
d
h
o
p
_
le
n
g
th
o
f
5
1
2
.
Nex
t,
th
e
FF
T
is
ap
p
lied
to
co
n
v
er
t
ea
ch
f
r
am
e
f
r
o
m
th
e
tim
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
la
s
s
i
fica
tio
n
o
f reg
io
n
a
l la
n
g
u
a
g
e
d
i
a
lects u
s
in
g
co
n
v
o
lu
tio
n
a
l n
eu
r
a
l
n
etw
o
r
k
…
(
F
a
h
mi
B
.
Ma
r
a
s
a
b
ess
y)
5023
d
o
m
ain
to
t
h
e
f
r
e
q
u
en
c
y
d
o
m
a
in
.
T
h
e
FF
T
v
alu
es
o
b
tai
n
ed
ar
e
u
s
ed
to
ca
lcu
late
s
p
ec
tr
al
en
e
r
g
y
d
e
n
s
ity
,
wh
ich
is
th
en
m
ap
p
ed
to
a
f
ilter
b
an
k
.
T
h
e
en
er
g
y
o
f
th
e
f
ilter
b
a
n
k
is
ca
lcu
lated
b
y
m
u
ltip
ly
in
g
th
e
en
er
g
y
s
p
ec
tr
u
m
with
th
e
f
ilter
b
a
n
k
,
th
en
s
u
m
m
in
g
t
h
e
c
o
ef
f
icien
ts
.
Af
te
r
th
e
f
ilter
b
an
k
is
co
m
p
u
ted
,
it
is
s
ca
led
to
t
h
e
Me
l
s
ca
le
to
o
b
tain
th
e
Me
l
-
s
ca
led
f
ilter
b
an
k
.
Fin
ally
,
a
D
C
T
is
ap
p
lied
to
th
e
s
ca
led
f
il
ter
b
an
k
,
r
etain
i
n
g
a
n
u
m
b
er
o
f
c
o
ef
f
icien
ts
an
d
d
is
ca
r
d
in
g
th
e
r
est.
T
h
e
f
in
al
r
es
u
lt
is
a
f
ea
tu
r
e
v
ec
t
o
r
with
4
0
MFC
C
co
ef
f
icien
ts
r
ep
r
esen
tin
g
th
e
au
d
io
in
f
o
r
m
atio
n
.
3
.
3
.
E
x
perim
ent
a
l
re
s
u
lt
s
W
ith
o
u
t
ap
p
ly
in
g
d
ata
a
u
g
m
en
tatio
n
tech
n
iq
u
es,
b
o
th
C
NN
an
d
ML
P
d
em
o
n
s
tr
ated
lim
ited
p
er
f
o
r
m
an
ce
,
with
ac
cu
r
ac
y
an
d
o
th
er
m
etr
ics
r
em
ain
in
g
s
u
b
o
p
tim
al.
T
h
e
ap
p
licatio
n
o
f
au
g
m
e
n
tatio
n
tech
n
iq
u
es,
s
u
ch
as
ad
d
in
g
n
o
i
s
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itch
s
h
if
tin
g
,
r
esu
lted
in
n
o
tab
le
im
p
r
o
v
em
en
ts
in
m
o
d
el
p
er
f
o
r
m
an
ce
,
p
ar
ticu
la
r
ly
f
o
r
t
h
e
C
NN
alg
o
r
ith
m
.
Mo
d
el
H,
wh
ich
co
m
b
in
es
all
th
r
ee
tech
n
iq
u
es,
ac
h
iev
ed
th
e
h
ig
h
est
ev
alu
atio
n
m
etr
ics
o
n
th
e
cu
r
r
e
n
t
d
ataset.
Ho
wev
er
,
wh
ile
th
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
co
m
b
in
in
g
au
g
m
en
tatio
n
tech
n
i
q
u
es
ca
n
en
h
a
n
ce
p
er
f
o
r
m
an
ce
,
th
eir
ef
f
ec
tiv
en
ess
m
a
y
d
e
p
e
n
d
o
n
th
e
d
ataset
ch
ar
ac
ter
is
tics
,
an
d
f
u
r
t
h
er
v
a
lid
atio
n
o
n
lar
g
er
,
m
o
r
e
d
iv
er
s
e
d
atasets
i
s
n
ec
ess
ar
y
to
co
n
f
ir
m
th
eir
g
en
er
al
ap
p
licab
ilit
y
.
Fo
r
th
e
ML
P
alg
o
r
ith
m
,
th
e
h
i
g
h
est
p
er
f
o
r
m
an
ce
was
ac
h
iev
ed
b
y
m
o
d
el
F,
w
h
ich
co
m
b
in
e
d
ad
d
n
o
is
e
an
d
p
itc
h
s
h
if
t
a
u
g
m
e
n
tatio
n
t
ec
h
n
iq
u
es.
B
ased
o
n
T
a
b
le
3
,
th
e
p
er
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
b
etwe
en
C
NN
an
d
ML
P
in
d
icate
s
th
at
th
e
ap
p
licatio
n
o
f
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es
s
ig
n
if
ican
tly
im
p
r
o
v
ed
b
o
th
m
o
d
els.
Ho
wev
er
,
th
e
d
if
f
er
in
g
r
esp
o
n
s
es
o
f
ea
ch
alg
o
r
ith
m
to
au
g
m
en
tatio
n
tech
n
iq
u
es
h
ig
h
li
g
h
t
th
e
im
p
o
r
tan
ce
o
f
s
elec
tin
g
an
d
ap
p
l
y
in
g
th
e
ap
p
r
o
p
r
iate
au
g
m
en
tatio
n
tech
n
iq
u
es tailo
r
ed
to
th
e
m
o
d
el’
s
r
e
q
u
ir
em
en
ts
.
Mo
d
el
H
d
em
o
n
s
tr
ated
th
e
h
i
g
h
est
ev
alu
atio
n
m
etr
ics
in
th
is
s
tu
d
y
f
o
r
r
eg
io
n
al
d
ialec
t
class
if
icatio
n
u
s
in
g
th
e
C
NN
alg
o
r
ith
m
.
B
y
ap
p
ly
i
n
g
d
ialec
t
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
co
m
b
in
in
g
all
th
r
ee
au
g
m
en
tatio
n
tech
n
iq
u
es,
th
e
m
o
d
el
u
tili
ze
d
3
,
8
2
4
tr
ain
in
g
s
am
p
les
an
d
2
4
0
test
in
g
s
am
p
les.
T
h
e
ev
alu
atio
n
m
etr
ics,
in
clu
d
in
g
an
ac
c
u
r
ac
y
o
f
9
7
.
9
2
%,
r
ec
all
o
f
9
7
.
9
%,
p
r
ec
is
io
n
o
f
9
7
.
9
7
%,
F1
-
s
co
r
e
o
f
9
7
.
9
2
%,
an
d
C
o
h
en
’
s
Kap
p
a
s
co
r
e
o
f
9
7
.
4
9
%,
s
u
g
g
est
th
at
th
e
m
o
d
el
p
er
f
o
r
m
s
w
ell
u
n
d
er
th
e
co
n
tr
o
lled
c
o
n
d
i
tio
n
s
o
f
t
h
is
s
tu
d
y
.
Ho
wev
er
,
g
iv
en
th
e
d
ataset'
s
lim
ited
s
ize
an
d
s
co
p
e,
f
u
r
t
h
er
r
esear
c
h
is
n
ee
d
ed
to
v
a
lid
ate
th
e
m
o
d
el'
s
ad
ap
tab
ilit
y
to
d
i
v
er
s
e
an
d
n
o
i
s
y
r
ea
l
-
wo
r
ld
co
n
d
itio
n
s
.
T
ab
le
3
.
R
ec
ap
o
f
p
er
f
o
r
m
a
n
c
e
m
etr
ic
co
m
p
a
r
is
o
n
b
etwe
e
n
C
NN
an
d
MLP
M
e
t
h
o
d
M
e
t
r
i
c
Ex
p
e
r
i
m
e
n
t
a
l
m
o
d
e
l
A
B
C
D
E
F
G
H
M
LP
A
c
c
u
r
a
c
y
2
7
.
9
2
9
5
.
8
3
9
6
.
6
7
3
0
.
8
3
9
6
.
6
7
9
7
.
5
0
9
5
.
4
2
9
6
.
6
7
R
e
c
a
l
l
2
6
.
6
5
9
5
.
9
1
9
6
.
4
4
2
9
.
6
7
9
6
.
4
4
9
7
.
6
4
9
5
.
0
5
9
6
.
6
5
P
r
e
c
i
s
i
o
n
2
3
.
5
1
9
6
.
0
1
9
7
.
2
1
2
8
.
5
1
9
7
.
1
4
9
7
.
5
0
9
6
.
3
9
9
6
.
8
8
F1
-
s
c
o
r
e
2
2
.
7
9
9
5
.
9
3
9
6
.
7
3
2
5
.
8
1
9
6
.
7
2
9
7
.
5
6
9
5
.
5
0
9
4
.
8
8
K
a
p
p
a
1
2
.
9
7
9
4
.
9
8
9
5
.
9
8
1
6
.
3
3
9
5
.
9
8
9
6
.
9
9
9
4
.
4
7
9
7
.
4
9
C
N
N
A
c
c
u
r
a
c
y
2
0
.
8
3
9
6
.
2
5
9
5
.
4
2
2
8
.
7
5
9
7
.
5
0
9
7
.
0
8
9
6
.
2
5
9
7
.
9
2
R
e
c
a
l
l
2
2
.
2
6
9
6
.
0
7
9
5
.
5
5
2
9
.
6
9
9
7
.
5
0
9
6
.
9
5
9
6
.
21
9
7
.
9
0
P
r
e
c
i
s
i
o
n
1
1
.
8
1
9
6
.
5
9
9
5
.
5
6
4
2
.
5
6
9
7
.
5
9
9
7
.
3
9
9
6
.
4
4
9
7
.
9
7
F1
-
s
c
o
r
e
1
3
.
8
3
9
6
.
2
7
9
5
.
5
5
2
4
.
5
7
9
7
.
5
4
9
7
.
5
6
9
6
.
31
9
7
.
9
2
K
a
p
p
a
0
5
.
5
5
9
5
.
4
8
9
4
.
4
8
1
5
.
2
6
9
6
.
9
9
9
6
.
9
9
9
5
.
48
9
7
.
4
9
3
.
4
.
Dis
cus
s
io
n
T
h
e
p
r
o
p
o
s
ed
class
if
icatio
n
m
o
d
el
u
s
es
th
e
C
NN
alg
o
r
ith
m
b
ased
o
n
th
e
m
o
d
el
H
e
x
p
er
im
en
t,
ap
p
ly
in
g
th
e
ad
d
in
g
n
o
is
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itch
s
h
if
tin
g
tech
n
iq
u
es
to
ad
d
n
ew
s
y
n
th
eti
c
d
ata
to
th
e
test
in
g
d
ata.
T
h
e
n
,
to
d
eter
m
in
e
th
e
ch
ar
ac
ter
is
tics
o
f
ea
ch
au
d
io
s
am
p
le,
d
ialec
t
lan
g
u
a
g
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
is
a
p
p
lied
.
T
h
e
ev
alu
atio
n
m
etr
ics
in
d
ica
te
p
r
o
m
is
in
g
p
e
r
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el,
with
an
a
cc
u
r
ac
y
o
f
9
7
.
9
2
%
f
o
r
d
ialec
t
class
if
icatio
n
o
n
th
e
cu
r
r
e
n
t
d
ataset.
T
h
e
r
ec
all
s
co
r
e
o
f
9
7
.
9
%
s
h
o
ws
th
at
t
h
e
m
o
d
el
ef
f
ec
tiv
el
y
id
en
tifie
s
m
o
s
t
p
o
s
itiv
e
ca
s
es,
wh
ile
th
e
p
r
ec
is
io
n
s
co
r
e
o
f
9
7
.
9
7
%
d
em
o
n
s
tr
ates
its
ab
ilit
y
to
m
in
im
ize
f
alse
p
o
s
itiv
es.
T
h
e
F1
-
s
co
r
e,
r
ea
c
h
in
g
9
7
.
9
2
%,
s
u
g
g
ests
a
b
alan
ce
b
etwe
en
p
r
ec
is
io
n
a
n
d
r
e
ca
ll,
an
d
a
C
o
h
en
’
s
Kap
p
a
s
co
r
e
o
f
9
7
.
4
9
%
r
ef
lect
s
s
u
b
s
tan
tial
ag
r
ee
m
en
t
b
etwe
en
p
r
ed
ictio
n
s
an
d
tr
u
e
lab
els.
Ho
wev
er
,
g
iv
en
t
h
e
r
elativ
ely
lim
ited
d
ataset
an
d
s
co
p
e
o
f
th
e
s
tu
d
y
(
s
ix
d
ialec
ts
an
d
less
th
an
2
,
0
0
0
s
am
p
les),
f
u
r
th
er
ev
al
u
atio
n
s
o
n
lar
g
e
r
an
d
m
o
r
e
d
iv
er
s
e
d
a
tasets
ar
e
n
ee
d
ed
to
c
o
n
f
ir
m
t
h
e
g
en
er
aliza
b
ilit
y
a
n
d
r
o
b
u
s
t
n
ess
o
f
th
e
m
o
d
el.
Ad
d
itio
n
ally
,
c
h
allen
g
es
s
u
c
h
as
h
an
d
lin
g
in
to
n
atio
n
v
ar
ia
b
ilit
y
an
d
r
ea
l
-
wo
r
ld
n
o
is
e
co
n
d
itio
n
s
r
em
ain
ar
ea
s
f
o
r
f
u
tu
r
e
im
p
r
o
v
e
m
en
t
.
T
h
e
p
e
r
f
o
r
m
an
ce
o
f
t
h
e
C
NN
m
o
d
el
in
t
h
is
r
esear
ch
was
c
o
m
p
ar
ed
with
th
at
o
f
p
r
e
v
io
u
s
r
esear
ch
co
n
d
u
cte
d
b
y
E
r
n
awa
ti
an
d
R
ian
a
[
1
6
]
.
T
h
eir
s
tu
d
y
u
tili
ze
d
a
n
1D
C
NN
co
m
b
in
ed
with
MFC
C
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
o
n
t
h
e
J
av
a
-
SED
d
ataset,
ac
h
iev
in
g
a
n
ac
c
u
r
ac
y
o
f
9
6
.
4
3
%.
I
n
co
n
t
r
ast,
th
is
r
esear
ch
ap
p
lied
th
e
s
am
e
1
D
C
NN
ar
ch
itectu
r
e
an
d
MFC
C
f
ea
tu
r
e
ex
tr
ac
tio
n
to
th
e
I
n
d
o
n
esia
n
d
ialec
ts
d
ataset,
wh
ich
co
n
s
is
ts
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
1
7
-
5
0
2
6
5024
m
u
ltip
le
r
eg
io
n
al
d
ialec
ts
.
As
a
r
esu
lt,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
ac
h
iev
ed
an
im
p
r
o
v
ed
ac
c
u
r
ac
y
o
f
9
7
.
9
2
%.
T
h
is
in
d
icate
s
th
at
th
e
ar
ch
itectu
r
e
an
d
m
eth
o
d
o
lo
g
y
ad
ap
ted
f
r
o
m
th
e
p
r
ev
io
u
s
s
tu
d
y
ar
e
h
i
g
h
ly
ef
f
ec
tiv
e
wh
e
n
ap
p
lied
to
a
b
r
o
a
d
er
an
d
m
o
r
e
d
iv
er
s
e
d
atas
et,
h
ig
h
lig
h
tin
g
t
h
e
r
o
b
u
s
tn
ess
an
d
g
en
e
r
aliza
b
i
lity
o
f
th
e
1
D
C
NN
m
o
d
el
in
class
if
y
in
g
r
eg
io
n
al
d
ialec
ts
.
Fu
r
th
er
co
m
p
ar
is
o
n
with
s
tu
d
ies
o
n
d
ialec
t
class
if
icatio
n
al
s
o
r
ein
f
o
r
ce
s
th
ese
f
in
d
i
n
g
s
.
A
s
tu
d
y
b
y
Am
an
i
et
a
l
.
[
2
4
]
a
p
p
lied
x
-
v
ec
to
r
s
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
co
m
b
in
e
d
with
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM)
,
ac
h
iev
in
g
an
ac
c
u
r
ac
y
o
f
8
7
%.
Alth
o
u
g
h
x
-
v
ec
to
r
s
a
r
e
a
n
ad
v
an
ce
d
tech
n
iq
u
e
in
s
p
ea
k
er
r
ec
o
g
n
itio
n
,
th
e
r
elian
ce
o
n
SVM
ap
p
ea
r
s
to
lim
it
class
if
icat
io
n
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
C
NN.
An
o
th
er
s
tu
d
y
b
y
Gh
a
f
o
o
r
et
al.
u
s
ed
MFC
C
f
ea
tu
r
es
with
1
D
C
NN
o
n
a
Ku
r
d
is
h
d
ialec
t
d
a
taset
,
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
9
5
.
5
3
%.
Similar
ly
,
Kar
im
et
a
l
.
[
2
5
]
a
d
o
p
te
d
a
m
o
r
e
co
m
p
lex
a
p
p
r
o
ac
h
b
y
co
m
b
in
in
g
MFC
C
,
Me
l
s
p
ec
tr
o
g
r
a
m
,
p
o
ly
-
f
ea
tu
r
e,
a
n
d
co
n
tr
ast as f
ea
tu
r
e
ex
tr
ac
tio
n
t
ec
h
n
iq
u
es with
1
D
C
NN
,
r
esu
ltin
g
in
an
ac
c
u
r
ac
y
o
f
9
6
.
5
%.
C
o
m
p
ar
ed
t
o
th
ese
s
tu
d
ies,
th
e
m
eth
o
d
p
r
o
p
o
s
ed
i
n
th
is
r
esear
ch
ac
h
iev
ed
th
e
h
ig
h
est
ac
cu
r
ac
y
at
9
7
.
9
2
%.
T
h
is
s
u
g
g
ests
th
at
th
e
co
m
b
in
atio
n
o
f
1
D
C
NN
an
d
MFC
C
f
ea
tu
r
es
is
n
o
t
o
n
l
y
ef
f
e
ctiv
e
b
u
t
also
r
o
b
u
s
t
in
h
an
d
lin
g
v
ar
iatio
n
s
with
in
t
h
e
I
n
d
o
n
esia
n
d
ialec
t
d
ataset
.
No
n
eth
eless
,
it
is
im
p
o
r
tan
t
t
o
n
o
te
th
at
d
if
f
er
e
n
ce
s
in
d
ataset
ch
ar
ac
ter
is
tics
,
s
ize
,
an
d
ev
alu
atio
n
p
r
o
to
co
ls
ac
r
o
s
s
s
tu
d
ies
m
ay
in
f
lu
en
ce
th
e
r
ep
o
r
ted
r
esu
lts
.
T
h
er
ef
o
r
e,
f
u
r
t
h
er
r
esear
ch
is
r
ec
o
m
m
en
d
ed
to
v
alid
ate
th
e
g
en
er
aliza
b
ilit
y
an
d
r
o
b
u
s
tn
ess
o
f
th
is
ap
p
r
o
ac
h
ac
r
o
s
s
d
if
f
er
en
t
d
atasets
an
d
r
ea
l
-
wo
r
ld
s
ce
n
ar
io
s
.
4.
CO
NCLU
SI
O
N
B
ased
o
n
th
e
r
esear
ch
c
o
n
d
u
cted
,
s
ev
e
r
al
co
n
clu
s
io
n
s
ca
n
b
e
d
r
awn
as
a
f
o
u
n
d
atio
n
f
o
r
r
ec
o
m
m
en
d
atio
n
s
an
d
f
u
tu
r
e
r
esear
ch
d
ev
elo
p
m
e
n
t.
T
h
e
u
s
e
o
f
d
ata
au
g
m
e
n
tatio
n
tech
n
iq
u
es
s
u
ch
as
ad
d
in
g
n
o
is
e,
tim
e
s
tr
etch
in
g
,
an
d
p
itch
s
h
if
tin
g
h
as
s
ig
n
if
ican
tly
i
m
p
r
o
v
e
d
th
e
p
e
r
f
o
r
m
an
ce
o
f
d
ialec
t
class
if
icatio
n
f
o
r
r
eg
i
o
n
al
lan
g
u
ag
es.
T
h
e
r
esear
ch
r
esu
lts
s
h
o
w
th
at
m
o
d
els
im
p
lem
en
tin
g
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es
co
n
s
is
ten
tly
r
ec
o
r
d
ed
ac
cu
r
ac
y
im
p
r
o
v
em
e
n
ts
o
n
th
e
cu
r
r
e
n
t
d
ataset,
with
C
NN
ac
h
iev
i
n
g
a
n
ac
c
u
r
ac
y
o
f
9
7
.
9
2
%
an
d
ML
P
r
ea
ch
in
g
9
7
.
5
%.
T
h
is
s
u
g
g
ests
th
at
d
ata
a
u
g
m
en
tatio
n
ca
n
en
h
an
ce
m
o
d
el
p
er
f
o
r
m
a
n
ce
f
o
r
d
ialec
t c
lass
if
icatio
n
in
co
n
tr
o
lled
co
n
d
itio
n
s
.
Featu
r
e
ex
tr
ac
tio
n
u
s
in
g
MFC
C
al
s
o
p
lay
ed
a
s
ig
n
if
ican
t r
o
le
in
ca
p
tu
r
in
g
th
e
u
n
iq
u
e
ac
o
u
s
tic
ch
ar
ac
ter
is
tics
o
f
ea
ch
d
ialec
t.
Ho
wev
er
,
f
u
r
th
e
r
r
esear
ch
is
n
ee
d
ed
to
ev
alu
ate
th
ese
tech
n
iq
u
es'
ef
f
ec
tiv
en
ess
in
m
o
r
e
co
m
p
lex
an
d
d
iv
er
s
e
d
atasets
,
esp
ec
ially
u
n
d
er
r
ea
l
-
wo
r
ld
co
n
d
itio
n
s
.
T
h
e
C
NN
an
d
ML
P
alg
o
r
ith
m
s
d
em
o
n
s
tr
ate
s
tr
o
n
g
p
er
f
o
r
m
an
ce
i
n
d
ialec
t
class
if
i
ca
t
io
n
,
with
C
NN
co
n
s
is
ten
tly
s
h
o
win
g
b
etter
p
e
r
f
o
r
m
a
n
ce
th
an
ML
P
in
ter
m
s
o
f
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
k
ap
p
a.
Ho
wev
er
,
th
er
e
ar
e
s
till
ch
allen
g
es
r
eg
ar
d
in
g
t
h
e
m
o
d
el'
s
a
b
ilit
y
to
g
en
e
r
alize
to
n
ew
d
a
ta,
esp
ec
ially
wh
en
f
ac
in
g
v
ar
iatio
n
s
in
d
ialec
t
in
to
n
atio
n
a
n
d
ex
p
r
ess
io
n
.
T
h
er
ef
o
r
e,
f
u
r
th
er
r
esear
c
h
is
n
ee
d
ed
to
o
p
tim
ize
t
h
e
m
o
d
el
to
b
e
m
o
r
e
ad
ap
tab
le
to
a
wid
er
r
an
g
e
o
f
d
ialec
t v
ar
iat
io
n
s
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
au
th
o
r
s
wo
u
ld
lik
e
to
ex
p
r
ess
th
eir
s
in
ce
r
e
g
r
atitu
d
e
to
all
r
esp
o
n
d
en
ts
wh
o
willin
g
ly
p
ar
ticip
ated
an
d
co
n
tr
ib
u
ted
th
eir
v
o
ice
r
ec
o
r
d
in
g
s
f
o
r
th
e
p
u
r
p
o
s
es
o
f
th
is
s
tu
d
y
.
T
h
eir
s
u
p
p
o
r
t
an
d
co
o
p
er
atio
n
wer
e
in
s
tr
u
m
en
tal
in
th
e
d
ata
co
llec
tio
n
p
r
o
ce
s
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
e
d
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Fah
m
i B.
Ma
r
asab
ess
y
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Dwiza
R
ian
a
✓
✓
✓
✓
✓
✓
✓
✓
Mu
ji E
r
n
awa
ti
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
la
s
s
i
fica
tio
n
o
f reg
io
n
a
l la
n
g
u
a
g
e
d
i
a
lects u
s
in
g
co
n
v
o
lu
tio
n
a
l n
eu
r
a
l
n
etw
o
r
k
…
(
F
a
h
mi
B
.
Ma
r
a
s
a
b
ess
y)
5025
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
W
e
h
av
e
o
b
tain
ed
in
f
o
r
m
ed
c
o
n
s
en
t f
r
o
m
all
in
d
iv
id
u
als in
c
lu
d
ed
in
t
h
is
s
tu
d
y
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
a
v
ailab
le
f
r
o
m
th
e
c
o
r
r
esp
o
n
d
in
g
au
t
h
o
r
,
[
DR
]
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
K
.
K
a
s
i
y
a
r
n
o
a
n
d
S
.
A
p
r
i
y
a
n
t
o
,
“
T
h
e
i
n
f
l
u
e
n
c
e
o
f
g
l
o
b
a
l
i
s
a
t
i
o
n
o
n
t
h
e
s
h
i
f
t
i
n
l
o
c
a
l
l
a
n
g
u
a
g
e
a
n
d
c
u
l
t
u
r
a
l
i
d
e
n
t
i
t
y
,
”
J
o
u
r
n
a
l
C
o
r
n
e
r
o
f
Ed
u
c
a
t
i
o
n
,
L
i
n
g
u
i
st
i
c
s
,
a
n
d
L
i
t
e
ra
t
u
re
,
v
o
l
.
4
,
n
o
.
3
,
p
p
.
3
7
2
–
3
8
3
,
F
e
b
.
2
0
2
5
,
d
o
i
:
1
0
.
5
4
0
1
2
/
j
c
e
l
l
.
v
4
i
3
.
4
3
5
.
[
2
]
A
.
F
.
A
j
i
e
t
a
l
.
,
“
O
n
e
c
o
u
n
t
r
y
,
7
0
0
+
l
a
n
g
u
a
g
e
s:
N
LP
c
h
a
l
l
e
n
g
e
s
f
o
r
u
n
d
e
r
r
e
p
r
e
se
n
t
e
d
l
a
n
g
u
a
g
e
s
a
n
d
d
i
a
l
e
c
t
s
i
n
I
n
d
o
n
e
si
a
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
6
0
t
h
A
n
n
u
a
l
Me
e
t
i
n
g
o
f
t
h
e
Asso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
,
2
0
2
2
,
p
p
.
7
2
2
6
–
7
2
4
9
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
2
.
a
c
l
-
l
o
n
g
.
5
0
0
.
[
3
]
N
.
V
.
W
i
l
m
o
t
,
M
.
V
i
g
i
e
r
,
a
n
d
K
.
H
u
mo
n
e
n
,
“
L
a
n
g
u
a
g
e
a
s
a
s
o
u
r
c
e
o
f
o
t
h
e
r
n
e
ss
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
r
o
ss
C
u
l
t
u
r
a
l
Ma
n
a
g
e
m
e
n
t
,
v
o
l
.
2
4
,
n
o
.
1
,
p
p
.
5
9
–
8
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
7
7
/
1
4
7
0
5
9
5
8
2
3
1
2
1
6
9
3
6
.
[
4
]
C
.
F
o
t
i
o
u
a
n
d
K
.
K
.
G
r
o
h
ma
n
n
,
“
A
smal
l
i
sl
a
n
d
w
i
t
h
b
i
g
d
i
f
f
e
r
e
n
c
e
s?
f
o
l
k
p
e
r
c
e
p
t
i
o
n
s
i
n
t
h
e
c
o
n
t
e
x
t
o
f
d
i
a
l
e
c
t
l
e
v
e
l
l
i
n
g
a
n
d
K
o
i
n
e
i
z
a
t
i
o
n
,
”
Fr
o
n
t
i
e
rs
i
n
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
6
,
p
p
.
1
–
1
9
,
J
a
n
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
8
9
/
f
c
o
mm
.
2
0
2
1
.
7
7
0
0
8
8
.
[
5
]
C
.
S
y
a
m,
S
.
S
e
l
i
,
a
n
d
W
.
J.
A
b
d
u
,
“
D
y
n
a
m
i
c
s
o
f
l
a
n
g
u
a
g
e
i
n
t
e
r
a
c
t
i
o
n
i
n
mu
l
t
i
c
u
l
t
u
r
a
l
u
r
b
a
n
c
o
m
mu
n
i
t
i
e
s:
a
n
a
l
y
s
i
s
o
f
so
c
i
o
-
c
u
l
t
u
r
a
l
l
i
n
g
u
i
s
t
i
c
e
n
v
i
r
o
n
me
n
t
,
”
S
o
c
i
e
t
y
,
v
o
l
.
1
1
,
n
o
.
2
,
p
p
.
5
7
5
–
5
8
8
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
0
1
9
/
so
c
i
e
t
y
.
v
1
1
i
2
.
6
2
8
.
[
6
]
A
.
M
u
s
y
a
f
f
a
a
n
d
L.
S
.
D
e
w
i
,
“
A
n
a
n
a
l
y
t
i
c
a
l
st
u
d
y
o
f
l
a
n
g
u
a
g
e
s
t
y
l
e
s
i
n
d
i
f
f
e
r
e
n
t
E
n
g
l
i
s
h
d
i
a
l
e
c
t
s,
”
J
u
rn
a
l
N
a
k
u
l
a
:
P
u
s
a
t
I
l
m
u
Pe
n
d
i
d
i
k
a
n
,
B
a
h
a
s
a
d
a
n
I
l
m
u
S
o
s
i
a
l
,
v
o
l
.
2
,
n
o
.
5
,
p
p
.
2
2
2
–
2
3
1
,
J
u
l
.
2
0
2
4
,
d
o
i
:
1
0
.
6
1
1
3
2
/
n
a
k
u
l
a
.
v
2
i
5
.
1
0
5
1
.
[
7
]
A
.
W
a
n
g
,
“
S
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
f
o
r
d
i
f
f
e
r
e
n
t
d
i
a
l
e
c
t
s
a
n
d
a
c
c
e
n
t
s,
”
I
T
M
We
b
o
f
C
o
n
f
e
r
e
n
c
e
s
,
v
o
l
.
7
3
,
F
e
b
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
5
1
/
i
t
m
c
o
n
f
/
2
0
2
5
7
3
0
2
0
1
1
.
[
8
]
Q
.
Li
,
Q
.
M
a
i
,
M
.
W
a
n
g
,
a
n
d
M
.
M
a
,
“
C
h
i
n
e
s
e
d
i
a
l
e
c
t
s
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
:
a
c
o
mp
r
e
h
e
n
s
i
v
e
su
r
v
e
y
,
”
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
v
i
e
w
,
v
o
l
.
5
7
,
n
o
.
2
,
p
p
.
1
–
3
9
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
0
2
3
-
1
0
6
6
8
-
0.
[
9
]
K
.
J.
G
h
a
f
o
o
r
,
K
.
M
.
H
.
R
a
w
f
,
A
.
O
.
A
b
d
u
l
r
a
h
m
a
n
,
a
n
d
S
.
H
.
T
a
h
e
r
,
“
K
u
r
d
i
s
h
d
i
a
l
e
c
t
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
1
D
C
N
N
,
”
Aro
-
T
h
e
S
c
i
e
n
t
i
f
i
c
J
o
u
rn
a
l
o
f
K
o
y
a
U
n
i
v
e
rsi
t
y
,
v
o
l
.
9
,
n
o
.
2
,
p
p
.
1
0
–
1
4
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
4
5
0
0
/
a
r
o
.
1
0
8
3
7
.
[
1
0
]
D
.
E
.
-
G
r
i
s
a
l
e
s
e
t
a
l
.
,
“
C
o
l
o
m
b
i
a
n
d
i
a
l
e
c
t
r
e
c
o
g
n
i
t
i
o
n
b
a
s
e
d
o
n
i
n
f
o
r
m
a
t
i
o
n
e
x
t
r
a
c
t
e
d
f
r
o
m
s
p
e
e
c
h
a
n
d
t
e
x
t
s
i
g
n
a
l
s
,
”
i
n
2
0
2
1
I
E
E
E
A
u
t
o
m
a
t
i
c
S
p
e
e
c
h
R
e
c
o
g
n
i
t
i
o
n
a
n
d
U
n
d
e
r
s
t
a
n
d
i
n
g
W
o
r
k
s
h
o
p
(
A
S
R
U
)
,
2
0
2
1
,
p
p
.
5
5
6
–
5
6
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
S
R
U
5
1
5
0
3
.
2
0
2
1
.
9
6
8
7
8
9
0
.
[
1
1
]
B
.
T
a
w
a
q
a
l
a
n
d
S
.
S
u
y
a
n
t
o
,
“
R
e
c
o
g
n
i
z
i
n
g
f
i
v
e
ma
j
o
r
d
i
a
l
e
c
t
s
i
n
I
n
d
o
n
e
s
i
a
b
a
s
e
d
o
n
M
F
C
C
a
n
d
D
R
N
N
,
”
J
o
u
rn
a
l
o
f
P
h
y
s
i
c
s
:
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s
,
v
o
l
.
1
8
4
4
,
n
o
.
1
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
1
8
4
4
/
1
/
0
1
2
0
0
3
.
[
1
2
]
K
.
N
u
g
r
o
h
o
,
E.
N
o
e
r
s
a
so
n
g
k
o
,
P
u
r
w
a
n
t
o
,
M
u
l
j
o
n
o
,
a
n
d
D
.
R
.
I
.
M
.
S
e
t
i
a
d
i
,
“
En
h
a
n
c
e
d
I
n
d
o
n
e
si
a
n
e
t
h
n
i
c
s
p
e
a
k
e
r
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
d
a
t
a
a
u
g
me
n
t
a
t
i
o
n
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
J
o
u
rn
a
l
o
f
K
i
n
g
S
a
u
d
U
n
i
v
e
rsi
t
y
-
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
3
4
,
n
o
.
7
,
p
p
.
4
3
7
5
–
4
3
8
4
,
Ju
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
k
s
u
c
i
.
2
0
2
1
.
0
4
.
0
0
2
.
[
1
3
]
E.
S
h
a
n
d
y
,
A
.
H
.
A
n
s
h
o
r
,
a
n
d
D
.
A
r
d
i
a
t
m
a
,
“
I
mp
l
e
me
n
t
a
t
i
o
n
o
f
d
a
t
a
m
i
n
i
n
g
f
o
r
s
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
S
u
n
d
a
n
e
se
d
i
a
l
e
c
t
u
si
n
g
K
N
N
m
e
t
h
o
d
w
i
t
h
M
F
C
C
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
N
e
t
w
o
rks
,
Arc
h
i
t
e
c
t
u
re
a
n
d
H
i
g
h
P
e
rf
o
rm
a
n
c
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
6
,
n
o
.
3
,
p
p
.
1
1
4
5
–
1
1
5
8
,
2
0
2
4
,
d
o
i
:
1
0
.
4
7
7
0
9
/
c
n
a
h
p
c
.
v
6
i
3
.
4
2
2
6
.
[
1
4
]
I
.
G
.
A
.
G
.
A
.
K
a
d
y
a
n
a
n
e
t
a
l
.
,
“
B
a
l
i
n
e
s
e
t
e
x
t
-
to
-
sp
e
e
c
h
d
a
t
a
s
e
t
a
s
d
i
g
i
t
a
l
c
u
l
t
u
r
a
l
h
e
r
i
t
a
g
e
,
”
D
a
t
a
i
n
Br
i
e
f
,
v
o
l
.
6
0
,
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
i
b
.
2
0
2
5
.
1
1
1
5
2
8
.
[
1
5
]
D
.
W
a
n
g
,
S
.
Y
e
,
X
.
H
u
,
S
.
L
i
,
a
n
d
X
.
X
u
,
“
A
n
e
n
d
-
to
-
e
n
d
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
s
y
st
e
m
w
i
t
h
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
f
r
o
m
a
mu
l
t
i
l
i
n
g
u
a
l
a
u
t
o
m
a
t
i
c
s
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
m
o
d
e
l
,
”
i
n
I
n
t
e
rsp
e
e
c
h
2
0
2
1
,
A
u
g
.
2
0
2
1
,
v
o
l
.
3
,
p
p
.
3
2
6
6
–
3
2
7
0
,
d
o
i
:
1
0
.
2
1
4
3
7
/
I
n
t
e
r
s
p
e
e
c
h
.
2
0
2
1
-
3
7
4
.
[
1
6
]
M
.
Er
n
a
w
a
t
i
a
n
d
D
.
R
i
a
n
a
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
h
u
ma
n
e
m
o
t
i
o
n
s
b
a
s
e
d
o
n
J
a
v
a
n
e
s
e
s
p
e
e
c
h
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
a
n
d
mu
l
t
i
l
a
y
e
r
p
e
r
c
e
p
t
r
o
n
,
”
I
n
d
o
n
e
si
a
n
J
o
u
r
n
a
l
o
f
El
e
c
t
ri
c
a
l
En
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
(
I
J
EEI)
,
v
o
l
.
1
2
,
n
o
.
1
,
p
p
.
1
0
1
–
1
1
2
,
M
a
r
.
2
0
2
4
,
d
o
i
:
1
0
.
5
2
5
4
9
/
i
j
e
e
i
.
v
1
2
i
1
.
5
3
4
3
.
[
1
7
]
A.
-
O
.
B
o
u
d
r
a
a
,
K
.
K
h
a
l
d
i
,
T.
C
h
o
n
a
v
e
l
,
M
.
T
.
H
.
-
A
l
o
u
a
n
e
,
a
n
d
A
.
K
o
m
a
t
y
,
“
A
u
d
i
o
c
o
d
i
n
g
v
i
a
E
M
D
,
”
D
i
g
i
t
a
l
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
,
v
o
l
.
1
0
4
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
s
p
.
2
0
2
0
.
1
0
2
7
7
0
.
[
1
8
]
B
.
A
.
A
l
sa
i
f
y
,
H
.
S
.
A
.
A
r
j
a
,
B
.
Y
.
M
a
a
y
a
h
,
a
n
d
M
.
M
.
A
l
-
T
a
w
e
e
l
,
“
A
d
a
t
a
set
f
o
r
v
o
i
c
e
-
b
a
s
e
d
h
u
m
a
n
i
d
e
n
t
i
t
y
r
e
c
o
g
n
i
t
i
o
n
,
”
D
a
t
a
i
n
Bri
e
f
,
v
o
l
.
4
2
,
J
u
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
i
b
.
2
0
2
2
.
1
0
8
0
7
0
.
[
1
9
]
H
.
A
.
A
l
s
a
y
a
d
i
,
A
.
A
.
A
b
d
e
l
h
a
mi
d
,
I
.
H
e
g
a
z
y
,
B
.
A
l
o
t
a
i
b
i
,
a
n
d
Z.
T
.
F
a
y
e
d
,
“
D
e
e
p
i
n
v
e
st
i
g
a
t
i
o
n
o
f
t
h
e
r
e
c
e
n
t
a
d
v
a
n
c
e
s
i
n
d
i
a
l
e
c
t
a
l
A
r
a
b
i
c
s
p
e
e
c
h
r
e
c
o
g
n
i
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
5
7
0
6
3
–
5
7
0
7
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
7
7
1
9
1
.
[
2
0
]
I
.
G
.
B
.
A
.
P
.
P
a
r
a
mi
t
h
a
,
H
.
B
.
K
u
s
n
a
w
a
n
,
a
n
d
M
.
Er
n
a
w
a
t
i
,
“
P
e
r
f
o
r
ma
n
c
e
c
o
m
p
a
r
i
so
n
o
f
d
e
e
p
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m
f
o
r
s
p
e
e
c
h
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
rm
a
t
i
c
s
En
g
i
n
e
e
ri
n
g
(
J
-
C
o
s
i
n
e
)
,
v
o
l
.
6
,
n
o
.
2
,
p
p
.
9
9
–
1
0
6
,
D
e
c
.
2
0
2
2
,
d
o
i
:
1
0
.
2
9
3
0
3
/
j
c
o
si
n
e
.
v
6
i
2
.
4
4
3
.
[
2
1
]
H
.
D
.
S
a
p
u
t
r
a
,
A
.
I
.
E.
Ef
e
n
d
i
,
E
.
R
u
d
i
n
i
,
D
.
R
i
a
n
a
,
a
n
d
A
.
S
.
H
e
w
i
z
,
“
H
e
p
a
t
i
t
i
s
p
r
e
d
i
c
t
i
o
n
u
s
i
n
g
K
-
N
N
,
n
a
i
v
e
B
a
y
e
s,
su
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
,
m
u
l
t
i
l
a
y
e
r
p
e
r
c
e
p
t
r
o
n
,
a
n
d
r
a
n
d
o
m
f
o
r
e
st
,
g
r
a
d
i
e
n
t
b
o
o
st
i
n
g
,
k
-
me
a
n
s
,
”
J
o
u
r
n
a
l
M
e
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
,
n
o
.
4
,
p
p
.
9
6
–
1
0
0
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
3
7
0
3
4
/
me
d
i
n
f
t
e
c
h
.
v
1
i
4
.
2
1
.
[
2
2
]
F
.
A
r
i
f
i
n
,
A
.
S
.
P
r
i
a
m
b
o
d
o
,
A
.
N
a
su
h
a
,
A
.
W
i
n
u
r
si
t
o
,
a
n
d
T.
S
.
G
u
n
a
w
a
n
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
Ja
v
a
n
e
s
e
s
p
e
e
c
h
e
m
o
t
i
o
n
d
a
t
a
b
a
s
e
(
Jav
a
-
S
ED
)
,
”
I
n
d
o
n
e
si
a
n
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
(
I
J
E
EI
)
,
v
o
l
.
1
0
,
n
o
.
3
,
p
p
.
5
8
4
–
5
9
1
,
2
0
2
2
,
d
o
i
:
1
0
.
5
2
5
4
9
/
i
j
e
e
i
.
v
1
0
i
3
.
3
8
8
8
.
[
2
3
]
C
.
N
u
g
r
a
h
a
a
n
d
S
.
H
a
d
i
a
n
t
i
,
“
G
l
a
u
c
o
ma
d
e
t
e
c
t
i
o
n
i
n
f
u
n
d
u
s
e
y
e
i
ma
g
e
s
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
m
e
t
h
o
d
w
i
t
h
v
i
s
u
a
l
g
e
o
me
t
r
i
c
g
r
o
u
p
1
6
a
n
d
r
e
s
i
d
u
a
l
n
e
t
w
o
r
k
5
0
a
r
c
h
i
t
e
c
t
u
r
e
,
”
J
o
u
r
n
a
l
M
e
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
,
n
o
.
2
,
p
p
.
3
6
–
4
1
,
Ju
n
.
2
0
2
3
,
d
o
i
:
1
0
.
3
7
0
3
4
/
me
d
i
n
f
t
e
c
h
.
v
1
i
2
.
7
.
[
2
4
]
A
.
A
man
i
,
M
.
M
o
h
a
m
ma
d
a
m
i
n
i
,
a
n
d
H
.
V
e
i
s
i
,
“
K
u
r
d
i
s
h
s
p
o
k
e
n
d
i
a
l
e
c
t
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
x
-
v
e
c
t
o
r
s
p
e
a
k
e
r
e
m
b
e
d
d
i
n
g
B
T
-
s
p
e
e
c
h
a
n
d
c
o
m
p
u
t
e
r
,
”
i
n
2
3
r
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
,
S
P
EC
O
M
2
0
2
1
,
2
0
2
1
,
p
p
.
5
0
–
57
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
8
7
8
0
2
-
3
_
5
.
[
2
5
]
S
.
H
.
T
.
K
a
r
i
m
,
K
.
J.
G
h
a
f
o
o
r
,
A
.
O
.
A
b
d
u
l
r
a
h
m
a
n
,
a
n
d
K
.
M
.
H
.
R
a
w
f
,
“
A
mu
l
t
i
-
f
e
a
t
u
r
e
f
u
si
o
n
a
p
p
r
o
a
c
h
f
o
r
d
i
a
l
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
u
si
n
g
1
D
C
N
N
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
n
I
n
f
o
rm
a
t
i
c
s
Vi
s
u
a
l
i
za
t
i
o
n
,
v
o
l
.
8
,
n
o
.
3
,
p
p
.
1
2
4
6
–
1
2
5
2
,
2
0
2
4
,
d
o
i
:
1
0
.
6
2
5
2
7
/
j
o
i
v
.
8
.
3
.
2
1
4
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
1
7
-
5
0
2
6
5026
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Fa
h
m
i
B.
M
a
r
a
sa
b
e
ss
y
,
S
.
T.
,
M.
K
o
m
.
is
a
re
c
e
n
t
m
a
ste
r'
s
g
ra
d
u
a
te
fr
o
m
th
e
F
a
c
u
lt
y
o
f
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
,
Un
i
v
e
rsitas
Nu
sa
M
a
n
d
iri
,
Ja
k
a
rta,
In
d
o
n
e
sia
,
c
u
rre
n
tl
y
se
rv
in
g
a
s
th
e
Ch
a
ir
p
e
rso
n
o
f
th
e
Alu
m
n
i
As
so
c
iatio
n
o
f
I
n
fo
rm
a
ti
c
s
En
g
i
n
e
e
rin
g
a
t
Un
iv
e
rsitas
P
a
n
c
a
sila.
Ac
ti
v
e
l
y
i
n
v
o
l
v
e
d
i
n
se
v
e
ra
l
h
o
b
b
y
-
b
a
se
d
c
o
m
m
u
n
it
ies
,
h
e
is
p
a
ss
io
n
a
te
a
b
o
u
t
tec
h
n
o
l
o
g
y
a
n
d
c
o
m
m
u
n
it
y
e
n
g
a
g
e
m
e
n
t.
Wi
t
h
a
k
e
e
n
in
tere
st
i
n
in
f
o
rm
a
ti
c
s
e
n
g
in
e
e
rin
g
,
h
e
is
n
o
w
f
o
c
u
sin
g
o
n
a
d
v
a
n
c
i
n
g
th
e
ir
e
x
p
e
rti
se
in
a
re
a
s
su
c
h
a
s
so
ftwa
re
d
e
v
e
lo
p
m
e
n
t
,
d
a
ta
a
n
a
ly
sis,
a
n
d
e
m
e
rg
i
n
g
tec
h
n
o
lo
g
i
e
s.
He
is
o
p
e
n
to
c
o
l
lab
o
ra
ti
v
e
o
p
p
o
rt
u
n
it
ies
in
re
se
a
rc
h
a
n
d
in
n
o
v
a
ti
o
n
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
1
4
2
2
0
0
1
1
@
n
u
sa
m
a
n
d
iri
.
a
c
.
id
.
Pro
f.
Dr
.
Ir
.
Dw
iza
Ria
n
a
,
S
.
S
i
,
M
.
M
.
,
M.
K
o
m
.
,
IPU,
ASEAN.
En
g
.
h
a
s
b
e
e
n
a
p
e
rm
a
n
e
n
t
lec
tu
re
r
a
t
t
h
e
F
a
c
u
lt
y
o
f
In
f
o
rm
a
ti
o
n
Tec
h
n
o
lo
g
y
si
n
c
e
2
0
0
3
,
a
n
d
is
c
u
rre
n
tl
y
a
lec
tu
re
r
in
th
e
Do
c
to
ra
l
P
ro
g
ra
m
in
In
fo
rm
a
ti
c
s
a
t
Un
iv
e
rsitas
Nu
sa
M
a
n
d
iri
.
Cu
rre
n
tl
y
sh
e
se
rv
e
s
a
s
Ch
a
n
c
e
ll
o
r
o
f
U
n
iv
e
rsit
a
s
Nu
sa
M
a
n
d
iri
.
Co
m
p
lete
d
Do
c
to
ra
l
E
d
u
c
a
ti
o
n
(S
3
)
i
n
th
e
El
e
c
tri
c
a
l
a
n
d
In
f
o
rm
a
ti
c
s
En
g
in
e
e
rin
g
S
tu
d
y
P
r
o
g
ra
m
a
t
t
h
e
Ba
n
d
u
n
g
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
(IT
B)
in
2
0
1
5
.
Ac
ti
v
e
o
n
th
e
DK
I
Ja
k
a
rta
P
ro
v
in
c
e
Ap
ti
k
o
m
A
d
v
i
so
ry
B
o
a
rd
,
a
s
P
r
o
v
i
n
c
ial
Ap
ti
k
o
m
Ad
v
iso
r
y
Bo
a
rd
Wes
t
Ja
v
a
,
Ce
n
tral
Ap
ti
k
o
m
M
a
n
a
g
e
m
e
n
t,
APT
IKO
M
Jo
u
rn
a
l
P
u
b
l
ish
i
n
g
Tea
m
,
a
s
a
d
m
in
istr
a
to
r
o
f
LAM
In
f
o
c
o
m
Di
v
isio
n
I,
M
a
n
a
g
e
m
e
n
t
o
f
th
e
As
so
c
iatio
n
o
f
In
d
o
n
e
sia
n
P
ri
v
a
te
Un
i
v
e
rsiti
e
s
(APT
IS
I)
Re
g
i
o
n
III
DK
I
Ja
k
a
rta
f
o
r
t
h
e
2
0
2
2
-
2
0
2
6
p
e
ri
o
d
,
a
n
d
a
s
Vic
e
Ch
a
ir
o
f
IEE
E,
Co
m
p
u
tati
o
n
a
l
I
n
telli
g
e
n
c
e
S
o
c
iety
,
I
n
d
o
n
e
sia
Ch
a
p
ter.
S
h
e
s
k
il
led
in
c
o
m
p
u
ter
sc
ien
c
e
,
ima
g
e
p
r
o
c
e
ss
in
g
,
d
a
ta
m
in
in
g
,
p
u
b
li
c
sp
e
a
k
i
n
g
,
m
a
n
a
g
e
m
e
n
t,
in
f
o
rm
a
ti
o
n
sy
ste
m
s,
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
wiz
a
@n
u
sa
m
a
n
d
iri
.
a
c
.
i
d
.
Mu
ji
Er
n
a
w
a
ti,
M.
K
o
m
.
is
a
lec
tu
re
r
a
t
Un
iv
e
rsitas
N
u
sa
M
a
n
d
iri
,
tea
c
h
in
g
in
th
e
In
fo
rm
a
ti
c
s
S
t
u
d
y
P
ro
g
ra
m
at
F
a
c
u
lt
y
o
f
In
f
o
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
.
Ad
d
it
i
o
n
a
ll
y
,
sh
e
is
p
a
r
t
o
f
t
h
e
Re
se
a
rc
h
a
n
d
C
o
m
m
u
n
it
y
S
e
rv
ice
In
stit
u
te
(LP
P
M
)
tea
m
a
t
Un
iv
e
rsitas
Nu
sa
M
a
n
d
iri
,
sp
e
c
ifi
c
a
ll
y
in
t
h
e
In
tellec
tu
a
l
P
r
o
p
e
rty
Ce
n
ter
a
n
d
I
n
fo
rm
a
ti
o
n
S
y
s
tem
s
u
n
it
.
S
h
e
c
o
m
p
lete
d
h
e
r
m
a
ste
r’s
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
(M
.
Ko
m
.
)
in
2
0
2
3
.
S
h
e
h
a
s
p
u
b
li
sh
e
d
se
v
e
ra
l
re
se
a
rc
h
p
a
p
e
rs
in
n
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s,
fo
c
u
si
n
g
o
n
m
a
c
h
in
e
lea
rn
in
g
a
n
d
d
e
e
p
lea
rn
in
g
.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
m
a
c
h
in
e
lea
rn
i
n
g
,
d
e
e
p
lea
rn
in
g
,
a
n
d
n
a
tu
ra
l
la
n
g
u
a
g
e
p
ro
c
e
ss
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
u
ji
.
m
e
i@n
u
sa
m
a
n
d
iri
.
a
c
.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.